Android just got a boost it didn’t need. RIM, already staggering after a 5% market-share drop over the last quarter and the Playbook debacle, has just…well, “shot itself through the foot” fails to convey quite the right sort of intensity. In fact it has executed a hitherto-unprecedented form of marketing suicide which can only be characterized as a “double-tap Osborne through the head”.
Full story at BlackBerry OS 7: How to Osborne your smartphone sales (hat tip to Ken Burnside for the link and the post title). It’s well written, worth reading, and poignant for me because, yes, I myself once owned an Osborne 1. In brief, RIM just murdered the sales prospects of all its existing hardware and software in favor of a new handset that won’t ship for a couple of months and a new OS version that will not be available as an over-the-air update to its existing customers.
A point I think the authors didn’t emphasize quite enough is how badly this move is going to cheese off RIM’s carrier partners. RIM has just hammered not just its own revenue but the revenue the carriers were expecting from OS 6 handsets they had in inventory, which will now likely go unsold as customers realize they’ll never be upgradable. RIM relies on its carrier partners not just as a sales conduit but for most of its marketing, as well. How eager do you suppose the carriers are going to be to continue that, now?
It’s truly odd how something about the smartphone business seems to produce epic attacks of suicidal stupidity at formerly well-run companies, and the parallel with Nokia is becoming ever more compelling in this case. RIM’s initial strategic blunder – choosing to fight Android rather than co-opt it – has been followed by an escalating series of screwups in their execution, like the bad joke that is the Playbook.
Following this one, my previous assessment that RIM might be able to fort up around its more inertia-ridden corporate customers for 5-9% of continuing market share is out the window. They just threw that prospect away. Now I think RIM’s got a year to live, tops, with their best case being a buyout by somebody with a use for their infrastructure.
“It’s truly odd how something about the smartphone business seems to produce epic attacks of suicidal stupidity at formerly well-run companies, and the parallel with Nokia is becoming ever more compelling in this case.”
It’s not just the smartphone business,,,when an executive is in panic mode, and the adrenaline levels go up, rational thinking is not an option. Anything starts looking like a potential savior…
…just picking the low-hanging fruit….
Similarly, who really wants to buy a Nokia handset right now, when the NoWin models are in the pipeline?
Eric: A point you failed to mention — in addition to making a non-backwards-compatible version of their OS, on this earnings call, they’ve also announced that the QNX replacement (which may break all compatibility again) is about a year away….
“The two most common elements in the universe are hydrogen and stupidity.” — Albert Einstein.
“Not necessarily in that order.” — Harlan Ellison
I’m always skeptical of the Osborne Effect, because we all know that whatever we buy this year won’t be as cool as next year’s model.
That said, this looks like it may be a classic example. This kind of specific announcement begs people to wait.
It looks like it’s chickens-without-heads time at a lot of the wireless manufacturers. First Nokia, now RIM.
“The beancounters tell me that we can’t afford to buy $7 worth of moon rocks, let alone $70,000,000 of them Bought ’em anyway. Ground ’em up, made ’em into a gel. And let me tell you something: ground up moon rocks are pure poison. I am deathly ill!”
Oh, and lest you still cling to the illusion that Android’s openness will break carrier lockdown on phone capability — I’ll just leave this here:
http://thisismynext.com/2011/05/02/verizons-removal-tethering-apps-android-market-shame-fcc-violation/
>Oh, and lest you still cling to the illusion that Android’s openness will break carrier lockdown on phone capability — I’ll just leave this here:
Disturbing if true – but what’s the evidence that Google is actively cooperating? It’s more than possible that the relevant filter bits live on the handset and that Google was not responsible for turning them on – in fact, that arrangement is exactly what the carriers would want, in order to prevent Google from acting against their control.
>Oh, and lest you still cling to the illusion that Android’s openness will break carrier lockdown on phone capability
I momentarily forgot the final rebuttal to this kind of poor-mouthing. The carrier lockdown is already broken, because CyanogenMod exists.
Back in the good ol’ days when “portable” meant “neither bolted to the desk, nor chained to an alarm system!” That thing weighed 24 lbs! The Sun Ultra 20 M2 sitting on the desk next to me only ways 10 lbs more than that.
For those too young to remember, esr is alluding to the ‘Osborne Effect.’ When Adam Osborne started showing off the Osborne 1’s successor, the Osborne Executive off to the trade press a year before it was ready for release, sales of the Osborne 1 stopped, orders got cancelled, etc., in anticipation of the new model. Osborne couldn’t sell the massive quantity of 1’s in stock and eventually had to file bankruptcy over the mess.
Friend of mine says this: Microsoft and RIM should start making Windows phone 7 blackberries. So you add the Blackberry’s great hardware and Microsoft expertise in making Mobile OS-es together.
Makes sense? I think it does.
Short term, this RIM debacle will probably help Apple as much (maybe more) than Android, if some of the corporate phone stats we saw previously were correct.
Still, part of Jobs’s genius has been to go the opposite direction from Osborne — to separate new-gadget lust into two parts — the never sated unrequited lust for some intangible new thing that will always be around the corner, vs. the fully realizable lust for something new and cool and tangible, and available now.
By not announcing the tangible until he’s ready to ship a couple of million units at reasonable prices, he sells when the lust is at its peak, before it can fade and mingle with all the competing lusts for other things, real and imaginary, that people yearn for.
Osborne’s lesson, that announcing a product you can’t ship will kill the sales of the products you can ship, is certainly not heeded often enough. But I’m not sure I’ve seen anybody, at least not in high-tech, execute on Jobs’s lesson.
This lesson, that there is a flip side to Osborne — that, if you have enough faith in your creation to stockpile it a bit to feed the initial feeding frenzy, you can actually sell much more than if you simply announce it when you are ready to start shipping it and stop shipping its predecessor, indicates a keen grasp of the American psyche, with its need for instant gratification and conspicuous consumption.
>Short term, this RIM debacle will probably help Apple as much (maybe more) than Android, if some of the corporate phone stats we saw previously were correct.
Unlikely. The quarter in which Rim dropped 5% was one in which Apple only gained 0.2%. Even given that the overall smartphone market market was expanding, it’s not easy to come up with a scenario that allows Apple to have picked up more than 50% of those refugees; the way I read the numbers, the overall customer base would had to have to have increased by a factor of more than 12.5, which we know it didn’t do.
OTOH, unlike Osborne, I don’t think RIM is even arguably the market technology leader – this may not hurt them much because anybody who cares/is paying attention to that is *already* moving away from Blackberries. For the full-on Osborne you need to kill your own shipping product with vaporware – this is redundant when you’re already behind what other vendors are already shipping.
@esr
> I momentarily forgot the final rebuttal to this kind of poor-mouthing. The carrier lockdown is already broken,
> because CyanogenMod exists.
Sadly, Verizon/Motorola are going overboard trying to prevent this. I know the newer Droid models are a severe PITA to root and get a custom firmware on; my roommate recently was given a Droid X and he has been hitting wall after wall with it. Plus, CyanogenMod only officially supports the original Droid.
But, your statement stands. It’s just worth noting how hard Verizon is fighting it.
If CyanogenMod ever becomes widespread enough to even register on the radar of ordinary cellphone users,, expect the carriers to become very interested — and start cancelling accounts/lines for terms-of-service violations: it is, after all, theft of service from the carrier’s perspective. Casting dragnets for this sort of thing isn’t even hard: sniffing user-agent strings will get the vast bulk of the offenders.
The relevant filter bits appear to live in the Android Market app, which sniffs the SIM card (or in-handset ID for CDMA devices) to determine which carrier you’re on and filters out blocked apps accordingly. I’m hearing reports of people yanking their AT&T sim cards and being able to consequently download AT&T-blocked apps over WiFi.
That Google even implemented this sort of functionality implies that they were an active participant.
@esr:
> Unlikely. The quarter in which Rim dropped 5% was one in which Apple only gained 0.2%.
The Good Technologies report that I referenced here:
http://esr.ibiblio.org/?p=3152#comment-305296
claimed that, for the kind of locked down corporate customers they measure, Apple outsold Android 62% to 38% last quarter. To the extent that we’re already down to the nub of RIM customers who have to be on RIM because that’s what their corporation lets them have, Apple seems to be the winner for conversions, because they’ve convinced a lot of corporate IT departments that they are better (more locked down) than Android.
But to the extent that there are non-corporate people who actually like their RIM phone and were waiting for a better RIM phone, sure, those are now Android’s upgrades. I don’t know the percentages, but I assumed that most people who hated their RIM phone and could bail already did.
OTOH, as I have speculated earlier, it may be that these ex-RIM customers are one of the few groups that is helping Apple to hang on to its market share. For example, if you look at the charts in the Good survey, AT&T really did sell a crapload of iPhone 3s last quarter. It makes sense, for corporate accounts, where they don’t change carrier very often if at all, for those BlackBerrys to be traded in for cheapish iPhones.
But I don’t think any of the surveys are asking “Did you buy a smartphone in the last quarter? If so, what kind did it replace?” so we may just be left to speculate on those percentages.
> The relevant filter bits appear to live in the Android Market app, which sniffs the SIM card (or in-handset ID for CDMA devices) to determine which carrier you’re on and filters out blocked apps accordingly. I’m hearing reports of people yanking their AT&T sim cards and being able to consequently download AT&T-blocked apps over WiFi.
>
> That Google even implemented this sort of functionality implies that they were an active participant.
I can think of several legimate reasons for implementing this functionality. What it comes down to is that Google implemented a flexible tool, and the carriers are choosing to use this tool to … further their business model. Color me unsurprised.
What I’m curious about is what the carriers are going to do about phones that are running Gingerbread or later that have teathering built in? My Nexus-S has teathering and T-Mobile hasn’t said word one about it.
>I can think of several legimate reasons for implementing this functionality. What it comes down to is that Google implemented a flexible tool, and the carriers are choosing to use this tool to … further their business model. Color me unsurprised.
That is precisely my take on the situation.
@Jeff Read:
> That Google even implemented this sort of functionality implies that they were an active participant.
I have no reason to disbelieve that google made this sort of compromise to get the carriers on-board. I’m not sure I believe, as a lot of news accounts seem to be implying, that it’s a particularly new compromise. More likely spelled out in reams of documents signed three years ago.
For what it’s worth.
I have a nexus s, which has tethering built in (or at least it came with it) A t-mobile account, on a family plan with my kids, unlimted text and data, limited minutes.
As of right now, I configured tethering, fired up my brand new nook color, connected and downloaded a book that I purchased yesterday.
Worked quite well and the first try, an insignificant amount of time to download ‘Busting Vegas’.
Looks to me like this one is another example of stupidity in marketing. The ones that really want it will be the ones that use lots of bandwidth and will make it work unless the carriers use some sort of deep packet filtering that can tell if a packet is getting passed on to a laptop. All they will be acomplishing is to annoy there high end customers without truly blocking them.
Jim
Leave tethering out, of course. Handset manufacturers do android builds, not Google — and the carriers have considerable influence over that. (My Hero had a Sprint Now Network startup screen flashed into the bootup sequence. It showed even after I rooted and put CM on.)
This is my “dream” scenario for what should happen.
1) RIM dies
2) Whoever picks up the pieces releases QNX as totally free software. In the future era of 100 core CPUs, uKernels and message passing will shine.
3) Google would start porting their UI into something like the D programming language. Roughly speaking they’d have to port some 2 million LOC of java code into D to produce native running UI. Android would be rid of one of its biggest shortcomings – the VM two-step execution of of their code.
4) D goes viral as a result. Both GNOME,and KDE kick off projects to migrate their desktops to D. This will ultimately land them 10x the number over the long term.
5) Apple, Microsoft and C++ would die as a result. Humanity would reap the benefits.
You should have had the Lazarus Long quote, “Never underestimate the power of human stupidity”, as the epigraph for this one.
Absolutely, but there’s a difference between saying that there’ll be a better product next year and that the company is end-of-lifing all the current ones. RIM should simply have kept its mouth shut about the apparent incompatibility between 7 and its older hardware.
uma: Why D? Why not Google Go, which supports message passing concurrency natively in the language? Something like this could possibly happen, depending on the outcome of the Oracle/Google suit.
Anyway, free software mu-kernels and soft-RT operating systems exist already; given some extra work they could easily become competitive wth QNX.
> Why not Google Go, which supports message passing concurrency natively in the language? Something like this could possibly happen, depending on the outcome of the Oracle/Google suit
I haven’t looked as much into google’s Go. I would guess though that the step which has to be taken by the average programmer going from Java, or C++ into D is smaller than into Go. I don’t believe Go has any OOP facilities. While I am no fan of OOP it all comes down to how easy the average programmer will be able to make the transition without radically altering the way think about solving problems.
> Anyway, free software mu-kernels and soft-RT operating systems exist already; given some extra work they could easily become competitive wth QNX.
I looked at this a while back. MINIX3 does not have threads. They are required by many high level apps to compile on top of a MINIX core.The other kernels (including mathematically proven ukernels) have been more in the hypervisor area. QNX is really the “quickest” path to the dream uKernel world.
>I don’t believe Go has any OOP facilities.
It’s got some. I don’t completely understand them yet, but it looks like they’ve done something like a Self- or JavaScript-like object-by-example system without built-in inheritance.
Announcing something will ship in June when it’s already May doesn’t seem comparable to what Osborne did – if they lose a month or so of sales, that’s not going to make or break the company. Giving developers some reason to stick around might be worth that tradeoff, assuming the developers can get early access now.
Speaking of which, you probably want to comment on this:
http://www.bing.com/community/site_blogs/b/search/archive/2011/05/03/microsoft-and-rim-helping-people-make-better-decisions-with-bing-on-blackberry.aspx
I usually don’t post in these threads, but I’m curious: is ESR making speculative bets in the smartphone sector? All this chatter is interesting enough, but if you really want to convince people you need to put your own cash on the line and make some money.
>is ESR making speculative bets in the smartphone sector?
No. But the reason is not lack of confidence in my predictions, which have been pretty accurate so far. It’s that I don’t understand financial mechanics, and don’t have a strong enough desire to make boatloads of money to learn them and then pay a lot of attention to that game.
Financial mechanics are all math. It’d probably be a cinch for someone of your IQ and mathematical ability to pick up.
“Financial mechanics are all math. It’d probably be a cinch for someone of your IQ and mathematical ability to pick up.”
Yeah…the quants at the big banking houses and AIG were mathematical geniuses, weren’ they?
Financial mechanics are all math. It’d probably be a cinch for someone of your IQ and mathematical ability to pick up.
Knowledge does not equal ability. As someone who once wrote software for fund managers and traders, I saw this disconnect between knowledge and ability all of the time.
My experience is that the vast majority of people want to be right, and that desire to be right greatly interferes with their ability to make money.
I don’t doubt that you have confidence in your predictions, just that it’s easy to produce lots of explanations and tell yourself how great your predictions are, but much harder to deliver the goods in some way. I’m not suggesting that just putting money up gives you more authority, but that if you can make relatively consistent profit by investing according to your predictions then that proves the value of your predictions, at least in the short term.
>I’m not suggesting that just putting money up gives you more authority, but that if you can make relatively consistent profit by investing according to your predictions then that proves the value of your predictions, at least in the short term.
You’re quite right, but you’ve misunderstood my incentives. Gearson-Lehman plays me $300 an hour for consulting with VCs and investment banks, and I get not just steady but repeat business. So the “money talks” metric already tells me I’m quite good at the predictive game. I have little enough interest in making money per se to not want to become a financial mechanic. So why else should I do it? Merely convincing people of my accuracy who won’t pay me $300 per hour for it seems pointless.
@Jeff Reed>> That Google even implemented this sort of functionality implies that they were an active participant.
>>
@Craig> I can think of several legimate reasons for implementing this functionality. What it comes down to is that Google
> implemented a flexible tool, and the carriers are choosing to use this tool to … further their business model. Color me
> unsurprised.
>
@esr> That is precisely my take on the situation.
You can still side-load WiFi-tether: http://code.google.com/p/android-wifi-tether/downloads/list
But quoth Andy Rubin (from 2009):
“We also look forward to the day when consumers can access any application, including VoIP apps, from any device, on any network.”
s/VoIP/tethering/g and you’ll see the point. Android was promised to be “open”, but is much less than open now that the carriers are pushed it to success.
@esr> Gearson-Lehman plays me $300 an hour
does it piss them off to have to correct your spelling? It’s Gerson Lehrman. Note the ‘r’.
http://www.glgroup.com/
Why post idle chatter on a blog? This whole thing would be much more interesting if the predictions were tested in a meaningful way. It’s not as though chatter about the future of the tech sector is hard to come by.
> Now I think RIM’s got a year to live, tops
No doubting they’re in big trouble but you might be underestimating the factors speaking in favour of a stay of execution, which include:
[I] Blackberry has a lot of brand loyalty. Three examples, which are unlikely to evaporate in 12 months:
-1- Blackberry’s product design & rollout part may be looking quite sad at the moment, but their back operations has the confidence of their buyers. Their refusal to deal with several government’s demands to open up their data to snooping stands out in a bleak landscape. For non-nerds, Blackberry is by far the best option for people who value privacy.
-2- Blackberry has the longest-established free mobile IM network (Blackberry Messenger) that I believe still sees by far the most mobile usage, and is still the only free mobile IM network that is usable without a data plan, a fact that drives a lot of sales in Europe.
-3- Blackberry Exchange Server is still popular with many large & middle-sized companies, many of whom do not permit non-RIM clients to connect.
So I don’t expect consumer demand to tank in an Osborne-like way.
[II] Distributors will be pissed off, but they might be loyal too, if they think that there is an end in sight. And distributors really, really don’t want a high proportion of their highest margin customers moving to Android, and not altogether delighted about them becoming Iphone customers. I think also RIM have a high proportion of sales through non-retail channels, which buys them time.
[III] Shareholders are cowards. A company the size of RIM is a difficult takeover target. Nothing will happen before RIM’s share price implodes, and I think that RIM shareholders have a high tolerance for bad news at this point.
My prediction: RIM will issue at least 6 profit warnings before it becomes clear what will happen to them.
Greed and stupidity have caused plenty of turmoil throughout history. I don’t expect that change any time real soon.
I wonder how much MSFT paid RIM to put Bing on their phones. Another one of those “you can’t make this stuff up” moments.
Confident that they are handing their data off to world governments to use in spying operations? Sure.
Europe doesn’t matter as much anymore. The days of Nokia being the largest mobile phone manufacturer by unit sales are over. The most influential markets are now North America and Asia Pacific.
Sure, you can use Blackberry’s mobile IM network without a data plan, but you still need Blackberry’s very expensive service. OTOH, with a standard unlimited data plan, you can use any number of free mobile IM networks on Android and iOS, including Google Talk/Google Voice, which means you can IM or voice with anyone with a gmail account.
The mid-market company I work for is planning to migrate their BES to Android and OWA. OWA comes free with exchange. BES doesn’t.
> It’s truly odd how something about the smartphone business seems to produce epic attacks of suicidal stupidity at formerly well-run companies, and the parallel with Nokia is becoming ever more compelling in this case.
I don’t think that the smartphone business is in any way special in this regard, except maybe in that the competition is moving faster than in most other areas. Large corporations doing stupid things is not exactly news.
You’ve stated several times that Nokia was “formerly well-run”. That covers maybe the early part of Jorma Ollila’s tenure as CEO, i.e. roughly the 90s, and possibly in some ways Kari Kairamo’s time before that. The rest of the 150+ years of Nokia’s history is quite checkered, to put it mildly. Even in the 90s, they were getting rid of a large number of unprofitable businesses as they focused on telecommunications. I guess that was the corporation being ‘run well’, even if it didn’t always quite look that way.
A completely off-topic piece of trivia: one of the more unique Nokia products was the control computer system for the Loviisa nuclear power plant built in the late 70s. The Finnish utility wanted to buy western reactors, but the bilateral trade agreements with the Soviet Union meant that the Soviets needed to export something to Finland in return for the Finnish goods (their money wasn’t much good). Mostly this meant oil, but also things like electric locomotives, and, as it happened, two nuclear reactor units. The utility bought two VVER pressurized water reactors (not the Chernobyl RBMK type) and the associated turbine plants from the Soviets, but a lot of the control systems etc. were from Siemens and the computers from Nokia. The Soviet design did not have a reactor containment that would pass western standards, and the primary coolant circuit was quite large in size because it wasn’t designed to fit in a containment structure. Ultimately the large-enough western-designed containment structures were built under a license from Westinghouse, leading the Loviisa project to be called ‘Eastinghouse’. The reason I wound up reading about all this was Fukushima, and it turns out that in the one-off ‘Eastinghouse’ design at Loviisa, the spent fuel pools are inside the containment, whereas the other two currently operating reactors in Finland at Olkiluoto were bought from the Swedish ASEA (incidentally now part of Westinghouse) and have the spent fuel pools outside of the containment just like the GE Mark 1 reactors at Fukushima. (The geologists tell us that happily the Baltic Sea is too shallow to produce a proper tsunami, so obviously nothing can possibly go wrong.)
You’re either kidding yourself or incredibly naïve if you think that other smartphone OS manufacturers (Google, in particular) aren’t doing the same.
Financial mechanics are all math. It’d probably be a cinch for someone of your IQ and mathematical ability to pick up.
Accounting is all basic arithmetic. The reason it’s a four-year course at college is that’s what it takes to learn what to add, subtract, multiply or divide. Similar for the financial mechanics behind betting in the stock market. The math is the easy part. Partially because the rest has lots of historically contingent accretions and exceptions which do not make for a logical, coherent whole.
If you’re gonna troll about spelling, it helps not to misspell my surname in the selfsame post.
AT&T can detect sideloaded tethering apps, and if you don’t volunteer to pay for their enhanced tethering plan, they will volunteer you instead.
As long as the four — wait, make that three — national carriers hold all the cards, forget about cellphone freedom or openness in the U.S.
To do speculative investing, you don’t need a four-year degree in accounting. You only need that if you’re going to be, well, an actual accountant. My bachelor’s degree is in business, with a concentration in information systems. Most colleges give you a two-semester class called something like “Financial Analysis for Managers” or “Financial Management” or some such, which is about all you need.
Once you understand the basic accounting equation (assets = liability + shareholder’s equity) the rest just sort of falls into place. Assets are anything you have of value, such as income, real property, equipment, etc., while liabilities are anything that costs you money, such as expenses, loan repayments, etc. Figuring out how much a company is worth is (mostly) about solving for shareholder’s equity.
All the numbers and ratios stock traders talk about that are discussed are simply used for analysis. Most of them aren’t very useful except for comparison with like companies in the same industry.
The trick to making good money in the market is to dp what Warren Buffet does: pick stocks that are priced less than their intrinsic value, which can only be computed through fundamental analysis rather than technical analysis. Fundamental analysis boils back to the aforementioned basic accounting equation and ignores the things of technical analysis (trading patterns and trends). And, as always, invest for the long-term, not the short-term. Call me silly, but Warren Buffet got very, very rich by doing exactly that.
@Morgan Graywolf: “The trick to making good money in the market is to dp what Warren Buffet does: pick stocks that are priced less than their intrinsic value, which can only be computed through fundamental analysis rather than technical analysis. Fundamental analysis boils back to the aforementioned basic accounting equation and ignores the things of technical analysis (trading patterns and trends). And, as always, invest for the long-term, not the short-term. Call me silly, but Warren Buffet got very, very rich by doing exactly that.”
But this isn’t nearly as easy as it was a few decades ago, because there are far more investors looking for these market inefficiencies, and they have much more sophisticated tools (e.g., software stock screens that quick identify good candidates for further research). I’m not saying it can’t be done, but it’s a lot of work and even dedicated professional value investors can struggle to find stocks that meet their criteria.
Morgan Greywolf wrote:
The trick to making good money in the market is to dp what Warren Buffet does: pick stocks that are priced less than their intrinsic value, which can only be computed through fundamental analysis rather than technical analysis.
I’ve worked for the financial sector, met my fair share of fund managers, and this is not the trick to making money.
My experience? The single most important thing is being extremely disciplined when it comes to selling.
You can make a ton of money with a stupid buying strategy, as long as your selling strategy was spot on. The converse is not true. Someone who is poor at selling will lose money in the long run, even if they have a good buying strategy.
Really, if you think Buffet buys and holds forever, then you have never looked at their annual reports or SEC filings. For example, during the 4th quarter, Berkshire Hathaway sold their position in Nike, Lowes, Bank of America, Comcast, and Nestle, among others. He actually sold more stocks than that, but those are the instantly recognizable names.
As per article :-
Anyone want to take bets on how well that will go for AT&T when they sign someone up to the tethering plan automagically who doesn’t use MyWi for tethering (but may have it installed)?
From reading comments:
So it appears they’re trying to socially engineer their way to fun and profit. You say you’re not and give some reason why your data usage is high. They’re hoping to catch those dumb enough to admit to it.
@esr – “Self- or JavaScript-like object-by-example system without built-in inheritance.”
I am sure you yourself know that, but for clarification to any readers of the comments, this sort “object-by-example” of language paradigm is called Prototype Inheritance ( http://en.wikipedia.org/wiki/Prototype-based_programming )
A different point of view on the Verizon tethering thing.
@Morgan
> Confident that they are handing their data off to world governments to use in spying operations? Sure.
Name a mobile data service provider that has a better track record. In fact, name any other that has any record of standing up to government spying requests.
> Europe doesn’t matter as much anymore.
I’m not talking about a World Domination scenario, but about a survival scenario. That RIM can point to increasing market share in the world’s most mature mobile communications market is something they can spin to investors.
> Blackberry’s very expensive service.
The voice plan typically costs around $10/month, and comes with unlimited free IM.
> The mid-market company I work for is planning to migrate their BES to Android and OWA. OWA comes free with exchange. BES doesn’t.
Yes. I expect BES to lose share to Android, The point is that for many companies, just being able to roll your own infrastructure doesn’t cut it. I think that Android will only be able to really hurt RIM here with new service provision. That, I expect, will take some time.
Enterprise services is fairly conservative. Microsoft’s server division is still making billions of profit after all these years of enterprise Linux.
@esr – regarding future openness that Android was supposed to bring, it would be highly educational to have your read on the following trends: http://www.mydroidworld.com/forums/android-news-rumors/6888-some-food-thought-bootloaders-rooting-manufacturers-carriers.html#post65013
>regarding future openness that Android was supposed to bring, it would be highly educational to have your read on the following trends
What’s this “was supposed to bring”? The openness is already here for a lot of us.
P3Droid’s dire predictions have not come true. In fact, Motorola says it will provide a bootloader unlock this year. Some of what P3Droid is worried about isn’t actually possible – detecting tethering, for example. The carrier can’t do that, per se; all it can do is note high data usage, ask you if you’re using a tethering app, and hope you’re stupid enough to admit it. Some other things he’s worried about would violate FCC regs – denying service to rooted phones, for example, would fall afoul of the conditions that Google triggered in the Block C auction.
It’s never wise to underestimate the degree of evil that cell carriers will attempt if they think they can get away with it. But their ability to get away with it is increasingly constrained. P3Droid worries about a conspiracy between the carriers and handset makers to enforce lockdown – but before you get too upset about this, notice that the handset makers have lots of incentives to defect. From their point of view, lockdown mechanisms are a cost on their engineering budget that drives sales to the unlocked competition, so a double lose.
> Why D? Why not Google Go […]?
Why either, when Google is developing PNaCL? (PNaCL == NaCL + LLVM)
@Jeff: sorry. I’ll get it right from now on.
The Smartphone Wars: The RIse of AndroidNoG
http://blog.omniar.com/post/5179418792/rafer-sez-welcome-to-the-future-on-the-left-is
Witness the $87 AndroidNoG phone: http://armdevices.net/2011/04/11/best-of-shenzhen-87-android-3-5-capacitive-phone-mtk6516-fg8/
In other words, what I’ve been saying for sometime now, on this blog and elsewhere.
Amazon and Facebook can create an Android device *without* Google. No, it won’t be called an Android phone, but it will instead be called something more powerful: Facebook Phone or Kindle Phone. Baidu will almost certainly do a better job of embedding its search onto “Android” devices in China than will Google.
Microsoft has as much money to play with as Google does. Nothing to stop them from cutting deals with ZTE, Huawei, or even LG to wipe Google clean from the latest Android – Google build and put Bing, Bing Maps, etc. on everything from high-end to dirt-cheap “Android” phones.
How much is Google going to have to pay us to use Android? It gives away the operating system. It gives away search, maps and numerous other features that they’ve optimized for the platform, all so we will cick on that ad or click to call. At what point, I wonder, does the cost of Android become greater than the return?
I predict Android (with Google) will fail, and this failure will be different from all other Google failed products. All the others have just been shut down, and after celebrating the failure, quickly forgotten. Not Android. Android will be Frankenstein’s monster. (Remember what happened to Dr. Frankenstein?)
Non-Android Android phones will make waves, kill dumbphones and force Google to clarify its stance on what “open” really means. This is the truth in what Eric is preaching. But Non-Android Android phones will also seriously harm Google.
See this Bill Gurley piece: The Freight Train That Is Android — http://bit.ly/mGEJ5r
Also, some commScore numbers: Apple iOS Platform Outreaches Android by 59 Percent in U.S — http://bit.ly/jAWr45
and In Europe, Apple iOS Ecosystem Twice the Size of Android — http://bit.ly/mze3HD
Also, money for app developers: Apple Will Take 76 Percent of App Market Revenues This Year — http://bit.ly/lcYcS6
In the Smartphone Wars, there are two races: one to the bottom, and one to the bank.
Android makes volume, Apple makes profits.
>Witness the $87 AndroidNoG phone: http://armdevices.net/2011/04/11/best-of-shenzhen-87-android-3-5-capacitive-phone-mtk6516-fg8/
Which actually runs Android Froyo, somewhat undermining your thesis.
But I’m not worried about AndroidNoG, and I’m pretty sure Google isn’t either. Remember, their grand strategy isn’t aimed at control, it’s aimed at preventing anyone else from taking control. Bill Gurley hasn’t got it quite right; Google isn’t playing the game he thinks it is, but a more subtle one instead.
As for Facebook and Amazon fielding NoG phones, I’ll believe it when I see it and not sooner. That would be a remarkably stupid stunt, when they can already field apps on Android phones without incurring an OS development cost and take advantage of Google Android’s positive brand image to boot. All their incentives are to join Google, not beat it.
You propose that Microsoft might cut deals with handset makers to induce them to ship NoG phones, but the economics are stacked against this. Microsoft would have to cover the handset makers’ entire NRE for OS development, plus a premium representing the marketing value of Google/Android co-branding, otherwise a deal like that would be a net loss for the handset maker. It would be real nice if Microsoft chose to pour that much money down a rathole, and Ballmer might just be stupid enough, but I don’t think we can count on this happy outcome.
The Osborne Effect, at least for Osborne the company, is a myth deliberately created by Osborne the man to cover a general failure to execute by the company. Wikipedia covers this to some extent and I remember a full and specific confession by the man in I think a Wall Street Journal article or OE-ED some time ago in the previous century.
Yeah, I don’t see how Amazon could possibly incur the costs required to build an Android-based tablet without a hint of Android branding and with their own custom UI on top of it. It’s not like they’re a tech giant like Barnes & Noble, who is currently selling the second most popular tablet in the world by doing precisely that.
To be clear, the Nook is a win for Android as a platform. But it also bears out Richard’s predictions.
>To be clear, the Nook is a win for Android as a platform. But it also bears out Richard’s predictions.
That’s just silly. The percentage of search marketshare lost to Nook by Google Android is zero. The impact on Google’s ability to deliver ads is likewise zero. The Nook harms Google not at all.
@Richard, Bryant:
> To be clear, the Nook is a win for Android as a platform. But it also bears out Richard’s predictions.
As I’ve said multiple times before in many different ways, and as ESR just said: Google’s “grand strategy isn’t aimed at control, it’s aimed at preventing anyone else from taking control.”
The point is they needed to prevent an Apple walled garden. They’ve succeeded wildly. To them, this isn’t Frankenstein’s monster, it’s a truly open market where their search services can compete on an equal footing.
Will some other companies try to use Android to create their own walled gardens? Indubitably. Will that hurt google? Nowhere near as much as if they hadn’t brought Android to market, and probably not at all.
I find it interesting that the exact same people will complain in one breath that google hasn’t figured out how to monetize Android so it’s not doing them any good, and then in the next breath that other companies will take Android away from google. Google makes money when people can get to its search and other services. Android facilitates this. Even when somebody like B&N tries to hide Android CyanogenMod unhides it and then B&N realizes that it’s actually a good thing that people know its device runs Android.
@Cathy
Agreed. I didn’t say it was easy. :)
@hsu:
I didn’t say that. I said he invests for the long term. In keeping with your selling discipline, which I agree with, Buffett and Berkshire Hathaway sell when the stock becomes overpriced, again, based on fundamental analysis.
The expensive service is the Blackberry mail service, not the voice plan.
Using OWA isn’t “rolling your own infrastructure.” OWA is something most organizations are already using. Android’s Exchange Sync uses OWA. They don’t need to provision anything. There is no missing functionality.
@esr:
> P3Droid worries about a conspiracy between the carriers and handset makers to enforce lockdown – but before you
> get too upset about this, notice that the handset makers have lots of incentives to defect.
A thought: what if it is not a conspiracy, but extortion? “Help us solve this problem or we will stop selling your phones.”
>A thought: what if it is not a conspiracy, but extortion? “Help us solve this problem or we will stop selling your phones.”
They still have lots of incentives to defect. Their sales-maximizing strategy would be to agree, then put in an attempt at locking that looks plausible to the carrier but fails. Oopsie! Since designing a device that can’t be rooted but can be updated over the air is genuinely difficult – in fact, in principle it’s impossible – deniability is easy.
In any case, the oncoming collapse of the contract system will put paid to the carriers’ ability to even try this one on. In the near future most phones will be sold through WalMart or the likes of Bed Bath & Beyond (already selling Android tablets). The carriers will be out of the handset-sales picture, and their ability to control phone firmware will evaporate.
@Morgan, @Cathy, @hsu:
Your discussion around investing brought this investment axiom to mind:
“In the market Bulls make money, Bears make money and the Pigs get slaughtered.”
>>To be clear, the Nook is a win for Android as a platform. But it also bears out Richard’s predictions.
>That’s just silly. The percentage of search marketshare lost to Nook by Google Android is zero. The impact on Google’s ability to >deliver ads is likewise zero. The Nook harms Google not at all.
My bet is that it’s a small net gain for Google. ‘The inverse tragedy of the commons effect.’
Possibly not so small.
Offtopic, but peripheral, since you’re writing on marketshare: Why did “World Domination 201” fail to happen? Is the proliferation of Android devices the surprise method by which Linux devices become common for end users?
>Why did “World Domination 201” fail to happen?
The easy hypothesis is that the community blew it by not getting its act together on usability and codecs, which is the failure mode we were worried about. But in fact I think what happened is a combination of bribery, FUD, and strongarm tactics by Microsoft – knowing what was at stake, they made a maximum effort to lock us out of desktop/laptop/netbook preinstalls and succeeded. Dammit.
>Is the proliferation of Android devices the surprise method by which Linux devices become common for end users?
Sure looks like it, doesn’t it? Most of the Linux machines in the world are now smartphones. Turns out we were right about a technology disruption being required, after all.
I think you could say most of the Linux machines in the world are embedded devices, but not necessarily smartphones.
How many Linux settop boxes, NAS appliances, routers, TVs, microwaves, washing machines, etc. are there?
Some reasons why I think Linux failed dominate the desktop:
a) X-window system. Something like Wayland should have happened 4 or 5 years ago and been mature at this stage
b) Lack of productive low-barrier languages for developing the high level layers like UI. GNOME is written in C. KDE in C++. The success of Java and obj-c in the mobile space further illustrates this point.
c) Lack of Ubuntu-like standardization since the earliest days.
(d) hardware support
(a) and (c) are being solved now. (b) remains to be solved. (d) is getting better.
uma:
All true, but I do agree with ESR’s speculation that some kind of anti-competitive lockout on netbooks was a major factor. A year or two ago there were $200 Linux machines on sale at retail stores (Toys-R-Us!); today the prices are $280, and they all come with Windows 7 pre-installed.
I wonder if the real loser there isn’t Intel; if x86 could have blocked the coming x86->ARM disruption if a Wintel->Lintel platform shift was already in progress?
Only partially relevant for this particular post, but definitely smart phone wars related:
Not sure how this has been missed to now, but a company called ZTE is now the no. 4 phone manufacturer in the world.
http://www.gsmarena.com/zte_becomes_worlds_4th_largest_manufacturer_android_is_top_os-news-2270.php
For comparison, Blackberry is no. 5, HTC is no. 8. Big gap between no. 1 and 2 to no. 3 (no 1 Nokia shipped 453 million units last year, Samsung shipped 280 million, LG shipped 120 million, ZTE shipped 53 million phones).
And ZTE is shipping Android which is not only non-customized Android, but Android without the usual Google apps installed, which means they are not restricted by branding agreements with what they can do with the phone:
http://blog.omniar.com/post/5179418792/rafer-sez-welcome-to-the-future-on-the-left-is
From that second link:
“The phone is easily Nexus One quality and has a BOM of USD60. It’s SRP is about USD150. The developing world often hits volume on phones at an SRP of USD70. That’s at most 18 months from now, and it might happen this calendar year.”
That gives concrete example to some of the predictions from esr. I don’t remember which old posts are most relevant, so how does current BOM cost of $60 and SRP of $150 line up with older predictions? If this phone was available in the US, I would probably shell out $150 for an unlocked Nexus One quality device. I definitely would get one if it were $80, and I’m really a phone cheapskate.
The picture in the second link has a ZTE phone sitting next to a Nexus S, and the screen looks a lot nicer on the Nexus S, but it is also $350 more expensive. $350 would go a long way toward helping me overlook a slightly dimmer screen.
— Chris Caudle
> All true, but I do agree with ESR’s speculation that some kind of anti-competitive lockout on netbooks was a major factor. A year or two ago there were $200 Linux machines on sale at retail stores (Toys-R-Us!); today the prices are $280, and they all come with Windows 7 pre-installed.
I totally agree. My reasons were limited to technical reasons.
I think those same $200 machines will be back soon – and running on superior ARM. Perhaps running a modified version of Honeycomb this time around. Having the mobile and desktop experience unified for the average user will be a great plus.. And will further accelerate the decline of microsoft.
@Chris Caudle:
What do you mean, “missed”? Others have commented on ZTE as well. Here are a few places where I did…
http://esr.ibiblio.org/?p=2898#comment-294294
http://esr.ibiblio.org/?p=2905#comment-294874
http://esr.ibiblio.org/?p=3071#comment-302462
http://esr.ibiblio.org/?p=3071#comment-302519
http://esr.ibiblio.org/?p=3103#comment-303400
http://esr.ibiblio.org/?p=3152#comment-305211
@uma:
Perhaps, but in order for Android to supplant Windows on the desktop we’ll need to see more content creation tools as opposed to content consumption tools.
To see what I mean, try creating a spreadsheet on an Android smartphone. Now, I’m going to tell you that things have greatly improved in the last week and half or so with the introduction of Google Docs for Android, but as good as Google Docs’ spreadsheet is, its ability to handle very complex spreadsheets is quite limited and you still have to store your spreadsheet on Google’s servers.
For the Toys R Us crowd, a $200 netbook will be okay, but for anyone who does professional-level computing, it might be okay for light use, but it’s not enough to replace their existing desktop or laptop.
> For the Toys R Us crowd, a $200 netbook will be okay, but for anyone who does professional-level computing, it might be okay for light use, but it’s not enough to replace their existing desktop or laptop.
Sure. But it would still create a major dent in MS dominance. Most people do not do professional level work on their desktops. They just consume media, browse the web, read and write e-mail, download their digital camera photos (these type of activities). Google docs is plenty sufficient for those people and if not there are (or there will be) android applications that they can use for their needs. Who knows, maybe if the dent is large enough MS may be forced to create versions of their office suite for android – market pressure if the market proves large enough.
If things go well for desktop android, desktop windows for home users should theoretically be reduced down to its single digit numbers that exist for windows mobile – ie for those who absolutely must have MS Office for home use.
@Morgan:
Voices by those in the MS camp for microsoft to “abandon” the consumer market.
http://www.winsupersite.com/article/microsoft-products/Microsoft-Should-Abandon-the-Consumer-Market
>> To be clear, the Nook is a win for Android as a platform. But it also bears out Richard’s predictions.
>
> That’s just silly. The percentage of search marketshare lost to Nook by Google Android is zero. The impact on Google’s
> ability to deliver ads is likewise zero. The Nook harms Google not at all.
Only because the Nook uses Google as a search engine. It could just as easily be Bing.
The fundamental thesis is that Microsoft spends $s to get AndroidNoG manufacturers to use Bing search, Bing maps, etc.
Something you’ve not really addressed.
> Which actually runs Android Froyo, somewhat undermining your thesis.
If you look closely, it’s actually running a ripped-off version of the HTC ‘Sense’ UI. Trust me, Eric, it doesn’t have the “Full Girlfriend/Google Experience”. It is an AndroidNoG phone.
But then, every Cyanogen release is also Android NoG. No Google apps, no FGE.
> But I’m not worried about AndroidNoG, and I’m pretty sure Google isn’t either. Remember, their grand strategy isn’t aimed at control, it’s aimed at preventing anyone else from taking control.
“You will be. You will be.” http://www.yodajeff.com/multimedia/sounds/episode5/afraid.wav
@uma
I don’t disagree with that guy’s conclusion but his logic reads suspiciously like “lets define consumers as people who don’t buy microsoft products… we shouldn’t sell to those people”.
The 15-25% of revenue from consumers stuff is an interesting read however. I’ve long suspected something along those lines.
I think if this post shows anything, it shows something we all have kind of seen coming: that the smartphone wars are becoming a 2 horse race between Apple and Android. The other players like RIM are moving closer to oblivion.
In other news, according to data from IDC, Apple’s share of the global market recently increased from 16.7% to 18.1%. Read about it right here:
http://www.marketwatch.com/story/apple-gains-more-share-in-smartphone-market-2011-05-05?link=MW_latest_news
No signs of disruptive collapse, at least not yet.
Also, speaking of horseraces, the Kentucky Derby is this Saturday. Raise a glass.
@Darrencardinal:
I think you meant 16.1% to 18.7%. That’s really good.
According to the IDC data (which is coming out at multiple sites now), Apple managed to ship more handsets last quarter than ZTE, and managed to grow marketshare at a higher rate than both Samsung and HTC. Samsung and HTC apparently together still edge out Apple, rising from 18.1% to 19.7%.
I think the traditional dumbphone market is very rapidly switching to smartphones, and Apple is currently kicking ass when it comes to conversions.
Actually he has… it might be a bit roundabout but let me elaborate.
It appears (and we may be wrong) that your position is that to make android worthwhile for google everyone using Android must be using google [apps.*]. Eric’s thesis is that Google’s perspective is if they hadn’t released Android then the number of people using google (with the exception of expert users hacking their phone to do things specifically unwanted by the manufacturer) would be effectively zero. Thus Android is worthwhile to Google not because everyone is using Google apps but because it doesn’t lock you into using a competitor.
This is probably also why you can search using Bing, Yahoo or Ask Jeeves out of the box. If a regular user can change from Google to Bing on a phone with default config, they can change from Bing to Google on a phone configured by someone antagonistic(for whatever motivation) to Google.
I hope that IDC’s numbers about what happened (as mentioned by DarrenCardinal) are better than its numbers about what’s going to happen. We all know about those numbers.
Anyway, Intel makes their own projections, and comparing Intel’s PC numbers to IDC’s smartphone numbers from last quarter, it appears that smartphone shipments are just now crossing over PC shipments. Obviously they will grow a lot faster, but there’s still a lot of life left in the PC market.
Speaking of questionable IDC forecasts, they are now projecting that ARM will take 13% of the “PC chip market.” But I didn’t see in the articles exactly what constitutes a PC. Apparently, it’s just something running Windows, and IDC makes its claim based mostly on the fact that Microsoft announced they will be shipping Windows on ARMs.
http://www.eweek.com/c/a/Desktops-and-Notebooks/ARM-Will-Grab-13-Percent-of-PC-Chip-Market-in-2015-IDC-272226/
http://www.thinq.co.uk/2011/5/5/idc-predicts-success-arm-desktop-pcs/
Although I have no idea what the numbers will be and defer to the “experts”, it’s pretty obvious that more people will be open to the idea of an ARM in their laptop and desktop and tiny SheevaPlug/PogoPlug/whatever server, maybe even HTPC if the nVidia video acceleration makes up for the lack of horsepower — not needing a fan would be nice.
But I do have one relevant opinion:
In endorsing ARM for PCs, Microsoft is hastening its own destruction. Somebody who buys a cheap new ARM PC because it runs Windows, and then realizes that (a) it runs like a dog; and (b) none of the applications he is used to running will work on it because they are all X86, will be in an excellent position to scrape Windows off the system and either load an OS from the same company that makes his smartphone OS, or perhaps load a relatively polished Linux distro like Ubuntu ARM:
http://ubuntu-install.blogspot.com/2011/05/new-arm-based-ubuntu-machine.html
Don’t you think that people still might use Windows on ARM PCs just due to inertia?
@Morgan: There seems to be a little bit of talking past each other between us. I restore some context, and I maybe should try and rephrase my general point: ESR, and other commentators like John Gruber, point to indubitably dire facts about RIM’s situation, but then seem to assume that RIM is the next Palm, with the matching glass chin. I think that if you look at RIM’s situation, it shows a much more resilient business, one that has room to try out new strategies and make mistakes. I don’t think there is any risk of RIM switching to the No-Win strategy (though I don’t rule out them offering a virtualised WP on top of their mobile QNX offering).
>>>>still the only free mobile IM network that is usable without a data plan, a fact that drives a lot of sales in Europe.
>>>Sure, you can use Blackberry’s mobile IM network without a data plan, but you still need Blackberry’s very expensive service.
>>The voice plan typically costs around $10/month, and comes with unlimited free IM.
>The expensive service is the Blackberry mail service, not the voice plan.
A recent FT article (i) gave figures showing that voice-only plans for Blackberry phones have been steadily growing in numbers over the past two years in Europe, and (ii) gave substantial context showing that these are driven by parsimonious European consumers finding Blackberry’s offer the cheapest texting platform. Europe is remarkable in that it contains several countries that have seen fast growth in market share for RIM.
>>Yes. I expect BES to lose share to Android, The point is that for many companies, just being able to roll your own infrastructure doesn’t cut it. I think that Android will only be able to really hurt RIM here with new service provision. That, I expect, will take some time.
>Using OWA isn’t “rolling your own infrastructure.” OWA is something most organizations are already using. Android’s Exchange Sync uses OWA. They don’t need to provision anything. There is no missing functionality.
How many IT directors that deploy BES think that dropping BES and asking employees to use an open Exchange gateway would lead to no loss of valued functionality? Depending on what they want, replacing BES functionality would, I think, involve substantial IT commitments in terms of either costly product lock-in or labour to maintain infrastructure.
To be clear: I think BES is at risk from some open source project coming along and developing something that ticks enough boxes to convince these IT directors to switch, but if there is some such project, I don’t know of it.
>Don’t you think that people still might use Windows on ARM PCs just due to inertia?
Nobody ever wanted Windows on anything but X86. The NT ports to MIPS, HPPA, DEC Alpha, and so forth were all total failures.
> they made a maximum effort to lock us out of desktop/laptop/netbook preinstalls and succeeded. Dammit.
Oh, please… DELL offered Linux as a pre-install option for years, and they never saw any significant uptake. I know you ‘d like to believe that Linux on the desktop failed because of some dastardly action by Microsoft, but the fact is that Linux just wasn’t good enough to unseat the monopoly. Maybe if you’d taken X11 and the gaggle of half-assed, amateur GUIs out behind the barn and shot them in the head in the late 90’s you could have made a dent.
>Oh, please… DELL offered Linux as a pre-install option for years, and they never saw any significant uptake.
Dell’s Linux “offerings” were a lot like the Vogon construction notice in Hitchhiker’s Guide to the Galaxy. For all the marketing and company support they got, they might as well have been on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard’. By an odd coincidence, market support dollars from Microsoft equaled Dell’s profit margin in several quarters while this was going on.
That’s because they didn’t really “want Windows” so much as the ability to run various programs written for Windows. Since those programs have to be recompiled to run on other machine languages, you can’t take that program you paid for years ago and run it on your new computer. You have to go back to the vendor and buy it again, if they’re even selling that program, much less a port of it for that new architecture.
@The Monster:
> You have to go back to the vendor and buy it again, if they’re even selling that program, much less a port of it for that new architecture.
Which is exactly why I postulate that, to the extent that a Windows port to ARM drives any machine sales, Microsoft is hastening its own destruction. Once someone finds out that, even though the box says “Windows” his programs don’t run, then he will go looking for equivalent programs that do run. That’s a disruption that Linux is finally in position to take advantage of, for quite a few major program categories. Sure, the programs may not be as elegant or nice to work with. But there are an awful lot of free ones.
Alternatively, Microsoft can/should support Dalvik. Then you can run the MS Office suite, plus whatever Android apps you can find. Not a terrible value proposition, but one that a lot of customers will find to be fungible with Linux and OOffice…
Which brings us to:
@Some Guy:
> The NT ports to MIPS, HPPA, DEC Alpha, and so forth were all total failures.
I’m not exactly sure about that. To the extent they poisoned the well for any other possible mass market operating systems on those architectures, those ports may well have done their true job admirably. The ARM port could wind up in the same category, or it could be different for a couple of reasons: first, that ready-made free competitors already exist that are much more user friendly than any Unix or VMS back in the day; and second, that Microsoft has shown its willingness to really drop price to keep Windows on anything small and X86. Certainly, they will likewise come in at a low price on ARM. It merely remains to be seen exactly how appealing Windows is on ARM vs competitors such as Android, Chrome, Ubuntu, WebOS, etc.
Waitasecond, Richard Thompson, your theory here is that Microsoft will switch from pushing Windows Phone 7 to pushing Android NoG, and that’s somehow worse for Google than if Android never existed?
If Android had never existed, Windows Phone 7 might actually have had a chance, as every handset maker that couldn’t develop its own OS would have considered it as a way to compete with iPhone. At which point Microsoft wouldn’t have to bribe anybody to go “NoG”; people would be paying Microsoft for a NoG-in-licensing-terms phone OS. Alternatively, iPhone (Blackberry, WebOS) could have wound up dominant, with its already-closed ecosystem, and Microsoft would just have to write a check to One Infinite Loop (or other central HQ) to lock out Google entirely.
Instead, in an AndroidNoG world, MS can exclude Google from some phones and tablets across a market of multiple vendors, by paying the specific maker/carrier, with the exclusion going away when MS’s money goes away.
The dark horse in the processor wars is Via. They are always a couple of fab generations behind, but their new Eden X2 is state of the art in terms of X86 software support, can beat Atom on lots of benchmarks, and fairly low power to boot. All Via needs now is a competitive fab process.
As Charlie Demerjian points out, if Samsung gets pissed enough at Apple, they could stop making CPUs for them:
http://semiaccurate.com/2011/04/19/apple-suing-samsung-is-incredibly-stupid/
What if Samsung stopped fabricating Apple CPUs, sold a lot more ARMs to other smartphone makers, and used some of the excess capacity to fab CPUs for Via in a competitive process? Samsung would then support both ARM and X86, Apple would be incredibly wounded for a few quarters, and Via would be flying really high…
> It merely remains to be seen exactly how appealing Windows is on ARM vs competitors such as Android, Chrome, Ubuntu, WebOS, etc.
What if it runs the same programs it is able to run on x86? What if it can use an emulator to do the job? I myself can spot two problems with that. It will be slow (Although user are used to slow Windows), and other OS can do the same. However, does that make inertia and familiarity irrelevant?
To be clear, by OWA I mean Outlook Web
AccessApp; it’s a web-based mail client that comes standard with Microsoft Exchange. I don’t know of any enterprise Exchange customers that are not using it.> Sure, the programs may not be as elegant or nice to work with. But there are an awful lot of free ones.
I think many of the free applications are more “elegant” in their UI (and ease to work with) than the feature-crammed apps out there. For some strange reason there seems to be a natural convergence on Fitz-law friendly interfaces in free software. Also, those interfaces don’t change radically with consecutive releases (ribbon anyone ?).
A lot of people also seem to prefer small and slim applications whenever they get the job done (e.g. Abiword and Gnumeric vs. Openoffice) and this is an area where free software shines. This is not limited to linux. You see the same phenomenon on the mac platform. The relative success of Apple’s very own office suite (Pages, Numbers etc) etc which is slimmer than MS Office seems to back this up too. There is also a slim Abiword-like word processor that is quite popular on mac (Bean).
BTW, Adobe lately announced that they will port their suite to linux. Who knows maybe a gigantic exodus of the graphic design industry from windows to linux is to follow. They wouldn’t be doing this if there was no “demand”.
@esr:
You sound bitter. Both Dell and HP offer servers in tower cases with no OS. These make very nice Linux workstations. As for laptops, there’s always System 76.
@FLMKane:
For a start, but is MS going to create an X86 emulator (or reuse an open source one?) The question primarily boils down to network effects. A lot probably has to happen before a “compatible” ARM desktop or laptop can show up at a pricepoint that makes it attractive. Especially with Via’s low power cheap offerings coming online. If Microsoft puts in the effort, they can help make it happen, but if they just make a token effort, will there even be enough ARM Windows machines around to entice somebody else to put in the emulation effort?
Do you a have link to the announcement? Adobe has been hostile to porting their applications to Linux for years.
FWIW, Photoshop, Dreamweaver and Illustrator all run just fine under Wine these days.
@Morgan:
http://www.omgubuntu.co.uk/2011/04/adobe-photoshop-for-linux-definitely-on-the-radar/
There is also another link on the Adobe website where the asked people to stop posting a “port to linux” among the wish list because it was on the to do list. Apparently, they have been flooded with these requests and decided to listen.
The problem with OWA is it’s a pain in the ass to use, like mid-90s webmail.
That is until you open it in Inbternet Explorer, at which point it MAGICALLY BECOMES USABLE.
Clearly it must be because of the advanced technology in Microsoft’s web platform, which is available nowhere else.
for laughs: http://www.bgr.com/2011/05/06/windows-phone-sales-to-top-android-in-2015-smartphone-share-of-global-phone-sales-to-double
New comscore report (installed US smartphone OS and total handsets by manufacturer) is out.
At least according to this, iPhone 4V didn’t move the needle at all last quarter.
MS, Palm, and RIM still bleeding, but while Apple moved from 25% to 25.5%, Android moved from 28.7% to 34.7%, to overtake RIM as number one smartphone OS in the US.
> For all the marketing and company support they got,
…which was commensurate with the level of customer interest.
It’s not Dell’s job to change the world. They’re in the business of selling whatever they can move in volume, and desktop linux just didn’t have what it takes.
@esr: Dell’s Linux “offerings” were a lot like the Vogon construction notice in Hitchhiker’s Guide to the Galaxy.
@morgan greywolf: “You sound bitter. Both Dell and HP offer servers in tower cases with no OS. These make very nice Linux workstations.”
That’s fine for techies like us, but if your goal is World Domination (and more important, avoiding being locked out by someone else’s monopoly), you can only achieve that through pre-installs. The average user just will not install his own OS, even if you make it easy.
I’m bitter about this too. It was very underhanded of Microsoft to set up the “Windows tax” situation, effectively collecting revenue even from those who don’t want their products.
@Jeff: “The problem with OWA is it’s a pain in the ass to use, like mid-90s webmail. That is until you open it in Inbternet Explorer, at which point it MAGICALLY BECOMES USABLE. Clearly it must be because of the advanced technology in Microsoft’s web platform, which is available nowhere else.”
I use it on Firefox/Ubuntu all the time and it works fine. Of course, you have to ignore the nasty warning message that claims your platform is not compatible. Once you get past that, no problem.
The problem with Linux on the desktop is very simple; Linux isn’t compatible-enough with existing PC desktop binaries. The same phenomenon is why, despite being backed by Microsoft, OS/2 1.x never went anywhere (the “penalty box”), NT took until XP to become a desktop OS, and Vista had such problems. The 64-bit transition was not stark enough to dislodge Microsoft any more than the 32-bit transition or the GUI was, or similar evolutions in the mainframe space were to dislodge IBM. The only time DOS-Windows will become vulnerable on the desktop is if/when Microsoft gives up on backwards compatibility, and WINE is the best available platform for running existing PC binaries . . . or the desktop dies as a category.
Which I wouldn’t expect, actually. The desktop, I believe, will survive in business settings, even as it’s displaced by the new “nanocomputer” (smartphone/netbook/tablet/PDA/etc. complex) running Android in the home and pocket. You’ll wind up with the island of IBM business mainframes and the island of Microsoft business desktop micros in a sea of Linux machines (in the supercomputer, minicomputer/server, and nanocomputer roles).
According to comscore report (which is some sort of moving average, but should at least reflect approximately half the new Verizon iPhones introduced in the quarter):
Total number of US Smartphone subscribers 13+ years of age increased by 9.5M handsets, from 63M to 72.5M units.
Number of people carrying Apple smartphones increased by 2.75M handsets, from 15.75M to 18.5M
Number of people carrying Android smartphones increased by 7M handsets from 18M to 25M.
Interestingly, the comscore increase reflects almost exactly half the number of Apple handset activations in the US in the quarter (5.5M), but you would think it would also reflect around half the Apple handsets sold over Christmas. I guess there are a lot of replacements, or something.
@Patrick:
I wonder how the refurbishments that we have been reading about factor into the iPhone data.
ie the 2.75M delta, how many of those are refurbs of handsets that would other have been obsoleted ?
@uma:
Yeah, I wonder that too. I’ve discussed it in comments on previous posts. It’s unfortunate that it’s hard to tease out matching numbers from comscore and activation data but I guess that curiosity is how comscore makes money by selling reports :-)
Semi OT: It’s not exactly SoC. $25 computer on a USB stick. That makes me believe those $29 Android tablets at CVS/Walgreen’s (that will sell for $25 at Walmart) aren’t all that far off.
I just checked the page Monster linked to, and it triggered a thought. Does anyone know if there has been any work on using old smartphones to build Beowulf or similar system? After all, they are more powerful than the computers used to build the original Beowulfs.
> Both Dell and HP offer servers in tower cases with no OS.
Not only that, they’ll also pre-install whatever you want pre-installed on any order over ten units. Send them a disk image, and it’s “whatever you want, Mr. Customer!”
If there were customers for desktop Linux, you’d better believe the Dells and HPs of the world would be all over it.
>I’m bitter about this too. It was very underhanded of Microsoft to set up the “Windows tax” situation, effectively collecting revenue even from those who don’t want their products.
As I recall, back in the late ’90s, there were a handful of vendors like Penguin Computing who didn’t charge the Microsoft tax. How did they do?
@Some Guy:
Not very well. The Windows Tax (originally the DOS tax) was highly anticompetitive, although the FTC declined to act on it. It was basically something like “You can pay $100/copy for DOS, but if you ship it on every single machine, it drops to $50/copy.”
This works great for a product that has over 50% market share, to keep it over 50% market share. It makes DOS (later Window) “free” to the customer, as opposed to any other OS he might want to put on the system.
If the FTC had acted on that when it started, it would probably have saved a lot of grief and the major antitrust lawsuit.
Asus Eee Pad Transformer “first impressions” and reviews:
http://www.pcworld.com/article/227289/first_impressions_asus_transformer_tablet.html
http://www.pcworld.com/article/227373/asus_eee_pad_transformer_tf101_review_innovative_design_makes_this_tablet_stand_out.html
http://www.torontosun.com/2011/05/05/new-tablet-could-put-laptops-out-to-pasture
Apparently they were too timid after seeing the Xoom and Samsung tablets bomb, and didn’t realize the pent-up demand for a good, appropriately-priced Android tablet, so they aren’t building enough of them yet:
http://www.coated.com/asus-says-eee-pad-transformer-shortages-due-to-high-demand/
>It was basically something like “You can pay $100/copy for DOS, but if you ship it on every single machine, it drops to $50/copy.”
Which wouldn’t have mattered to a vendor that wasn’t selling Windows, would it?
> If the FTC had acted on that when it started, it would probably have saved a lot of grief and the major antitrust lawsuit.
Mostly it would have sold a few more copies of DR-DOS in 1992-1994 (when MS began really pushing “per-processor” and “per-system” licenses instead of per-copy), then choked off as MS unified DOS and Windows as a product.
@Steven Ehrbar:
“Mostly it would have sold a few more copies of DR-DOS in 1992-1994 (when MS began really pushing “per-processor” and “per-system” licenses instead of per-copy), then choked off as MS unified DOS and Windows as a product.”
As Eric already wrote, MS’ marketing program dollars were Dell’s margin. This per processor marketing program is the difference between profit and loss.
Note that MS have been repeatedly convicted for the practice of retaliatory pricing and that it is illegal to do this. So MS have time an again broken the law.
MS made a pledge to not level per processing pricing anymore, but they still do it in a roundabout way. You are saying that MS is willingly breaking the law for decades, and paying hundreds of millions of dollars in fines and settlement, just to execute a policy that is not working? You must know things MS never understood? However, I think it is more about blaming the victim.
Below is a listing from the European Committee for Interoperable Systems (http://www.ecis.eu/).
It is mostly notable for the references to the court cases, settlements, and convictions about MS’ illegal practices. So after two decades of $1B/year in legal fines and settlements [t]he only real difference between Microsoft’s more recent practices and its earlier ones is that, as Mr. Gates predicted, Microsoft has now changed its document retention practices.
Microsoft
A History of Anticompetitive Behavior and Consumer Harm
http://www.ecis.eu/documents/Finalversion_Consumerchoicepaper.pdf
And this is how it works on even the biggest companies:
@Some Guy:
Now that the internet has made commerce a lot more frictionless that statement is somewhat more true. But even now it’s pretty ridiculous, except for a really small, internet-only, SEO-savvy business. If 80% of your potential customers want Windows, deciding not to sell Windows is stupid.
Winter:
“Now the circle is complete!” (in Darth Vader voice)
> If 80% of your potential customers want Windows, deciding not to sell Windows is stupid.
I forgot to mention that this particular insidious anticompetitive practice is only available in the world of software. Physical objects don’t have the infinite gross margin that allow you to play these sorts of games.
As we found when Linux netbooks became plentiful and then disappeared, even free can have a hard time competing against an entrenched monopolist with fully amortized development costs who is willing to play this sort of game. There’s no doubt that the viable existence of Linux on netbooks reduced the end-user price of netbooks and low-end laptops considerably, by forcing Microsoft to drop the price, but it took a completely new market like smartphones to give Linux (Android) a serious chance at competing for the volume market. It will be really be interesting to see if this translates into any sort of desktop/laptop/notebook gains.
BTW, although Dell is certainly interested in Microsoft’s “marketing dollars,” those only go so far, and he actually does have to sell some stuff to make money. So, he’s more interested in beating Apple than in partnering with Microsoft. Which, these days, means Android:
http://news.cnet.com/8301-13506_3-20057010-17.html
@Patrick:
We need the ARM-based tablets to become a bridge to ARM-based laptops. We need to have that transition happen among the big manufacturers like Dell. Tablets have pretty much moved away from the idea of using Atom and x86. OEMs might be in a better shape to push linux back into marketplace on strictly-ARM designs. Microsoft will probably have less leverage to twist their arm if the linux offerings were all ARM. The OEMs will have a good story for Microsoft also that on x86 they enjoy their exclusive monopoly.
Microsoft porting to ARM is in reality a nail in their coffin.
@Patrick
The Asus tablet that you posted might be the beginning of that bridge.
@Jeff Read:
“The tighter Microsoft grips their OEMs, the more they will ooze through their fingers”…
Ewww.
@Patrick Maupin:
And, more importantly, we won’t be seeing Microsoft dominate the smartphone market mostly because the smartphone OEMs have Microsoft’s number already and they will not allow a repeat. For example, HTC’s chairwoman, Cher Wang, also helped to found VIA Technologies. She’s a smart cookie and she knows all about Microsoft’s game. HTC isn’t taking chances; rumor on the street is that they are working on their own Linux-based mobile OS to compete with Android. This is not a company with all their eggs in Google’s or anyone else’s basket. Might be interesting.
@Winter
I made the limited claim that the actual effects of earlier FTC action on per-processor licensing would have been restricted to benefiting DR-DOS in the short term. You’ve gone ahead and actually proved my point, by detailing the tactics available to and used by MS after the no-per-processor-licensing pledge. Insofar as they were illegal, you only have emphasized that that sort of minor regulatory effort would have had little effect, since Microsoft would actively subvert those efforts even if it meant breaking the law. So, I’ll say it again, all earlier FTC action on per-processor licensing would have done is sell a few more copies of DR-DOS before Microsoft adopted new tactics. If you think that statement is actually erroneous, please explain how.
And now I believe you owe me an apology for claiming my statement was “about blaming the victim.” If not, please explain exactly why you think a statement of belief that a specific regulatory intervention would have been ineffective is the equivalent as saying the victims of a lawbreaking organization were responsible for the results of the lawbreaking.
> If 80% of your potential customers want Windows, deciding not to sell Windows is stupid.
So, you’ve just said that trying to sell PCs with Linux instead of Windows is stupid. I agree completely.
>So, you’ve just said that trying to sell PCs with Linux instead of Windows is stupid. I agree completely.
Some Guy, you are trolling. And you are not being amusing. Stop now.
“And now I believe you owe me an apology for claiming my statement was “about blaming the victim.” If not, please explain exactly why you think a statement of belief that a specific regulatory intervention would have been ineffective is the equivalent as saying the victims of a lawbreaking organization were responsible for the results of the lawbreaking.”
That was not what I understood, so it was a genuine misunderstanding. And you did not blame the victim as I thought. My appologies for this accusation.
As far as regulatory intervention is concerned. The EU case of Samba vs MS showed that successful regulatory intervention is entirely feasible. However, the focus should be to protect the free choice of the consumer, not to redress some company’s losses.
> So, you’ve just said that trying to sell PCs with Linux instead of Windows is stupid. I agree completely.
If you make boxes, your job is to sell boxes to all those who need them. The more boxes you make (volume), the cheaper the boxes become (economy of scale) and the more you can grow your market share. You cannot lock yourself out of 80% of the market for the exact same box and compete.
What is so difficult about about this picture to comprehend ?
The only regulatory solution that would work is to prevent pre-installs.
The burden of supporting the zillion Dell, HP, Asus etc models would then shift to MS. They can can create a windows store. Whoever wants their OS can buy it for the machine they’re buying. Whoever wants something else can use that something.
That to me is competitive business environment. It would require the buyer to go MS’s site, and copy paste the Dell machine model that bought after the ordered their dell online. MS would ship them the OS image they need. It would take 20 more mins on behalf of each user to install that image. If you buy from Best Buy, best buy can install it for you for a nominal charge.
This type of regulation might be feasible in Europe. They regulate everything from the size of tomatoes to the number of scales that can be present on fish in their fishmarkets. Surely, they can surely erect this kind of barrier for the benefit of mankind.
@Some Guy:
> So, you’ve just said that trying to sell PCs with Linux instead of Windows is stupid. I agree completely.
I think we both agree that exclusively selling retail PCs with Linux instead of Windows is probably suicidal, except in special cases for special businesses. And as long as Microsoft, with their “marketing dollars”, can punish vendors who dare to sell Linux in addition to Windows, we will never know what the marketplace would think of stores that sell both.
esr:
I’m sure you’ve seen:
http://www.linuxfordevices.com/c/a/News/IDC-Quarterly-Smartphone-Tracker-1Q11/
@Jeff Read:
Not with Android it isn’t. With Android, you use the standard Android mail app and it uses OWA to sync with the Exchange server.
> Not with Android it isn’t.
It’s ridiculously easy. The hardest part is remembering to change my password on the phone when (correctly) forced to by corporate policies on my work laptop.
Yours,
Tom
Morgan,
I stand corrected. I’ve never had to connect to Outhouse with an Android phone.
Eric, do you have a comment on the rumor that Windows 8 will run on tablets? Do they have a chance of making a dent in the tablet space?
>Eric, do you have a comment on the rumor that Windows 8 will run on tablets? Do they have a chance of making a dent in the tablet space?
If they were able to ship today, yeah. But they’re probably eighteen months minimum from being able to ship that, and by that time they’ll have a hell of a hill to climb. Their record with WP7 doesn’t inspire optimism about their ability to attract app developers in competition with Android and Apple.
JB: I think moving Windows 8 to ARM is more about cheap servers than consumer tablets. Right now the biggest cost on the server room RoE isn’t hardware. It isn’t labor costs. It isn’t software licenses. It’s CPU cycles per watt of electricity.
Windows has been run on tablets for 10 years. If it was going to make a dent it would have already.
@JonB
I think people are way beyond the point where they only endure Windows and feel they have to fight Windows to use their computer.
It is long ago that I met people that actually liked to use Windows.
@Ken Burnside:
Several companies think they are going to make a lot of money selling ARMs into the server space for this reason. I could believe that current ARM implementations do better than current X86 implementations, but…
(1) The only independent organization that I know of doing serious work in server side benchmarks is SPEC, and they’ve had a power benchmark available since late 2007, and nobody has bothered to submit any ARM benchmark results. Occam’s Razor says that (a) the ARM-server wannabes don’t know how important this benchmark is to actual customers, which means they don’t understand the potential customers, which means they aren’t yet serious contenders; or (b) they do understand the importance of the benchmarks, but don’t want to submit any because they don’t yet measure up (but of course are ever optimistic that they will measure up soon).
(2) Even though Google bought Agnilux, a secretive hardware startup last year, and may actually be considering making silicon for data centers (or phones or whatever, who knows?), the view of Google’s senior vice president of operations seems to be that single-threaded performance still really matters a lot:
http://www.xbitlabs.com/news/cpu/display/20110301215553_Intel_Google_Doubt_ARM_and_Atom_Have_Chances_in_Servers.html
>Occam’s Razor says that
…the SPEC benchmarks are unimportant; the ARM vendors aren’t submitting results because they have too little visibility to the target market.
(I lost the point of the preceeding comment, cont.)
So it is understandable people do not choose Windows for a fun product like a tablet. That would just keep reminding them of their dayly struggle with firewalls and AV products.
You do not want to update your AV scanner on your shiny tablet every day? That is work.
@esr:
I’m confused about whether you are agreeing, disagreeing, or making a subtly different point. Certainly for the major datacenter customers that the ARM server-chip vendors like calxeda need in order to ramp volume, power benchmarks are extremely important as an evaluation starting point.
Calxeda claims that they can beat Intel performance per watt by 5x or 10x, even figuring in expected Intel advances.
In any case, if everything goes according to plan, there should be public knowledge of Calxeda’s performance by the end of the year.
Of course, we might not see Intel’s reply right away. I wouldn’t count them out after the first blow; certainly ARM itself (which is also one of Calxeda’s major investors) won’t.
>I’m confused about whether you are agreeing, disagreeing, or making a subtly different point.
I think the least hypothesis for why there aren’t SPEC ARM benchmarks is that the ARM vendors don’t think their potential customers care about SPEC benchmarks. I was pointing this out because I think you’re piling up too much conjecture on top of their absence.
Thanks. Your opinion matches my entirely uneducated opinion. Fall of 2012 seems like forever. There’s no way to get inside Android’s OODA loop as far as features at such a glacial pace.
The architectures of the future in the server game are ARM and Tilera style mesh architectures.
I have read somewhere that intel has internally produced non-x86 internal CPUs that are very similar to the Tilera-mesh architecture.
> the ARM vendors don’t think their potential customers care about SPEC benchmarks.
I see. I guess I conflated two different thoughts in my (a): “the ARM-server wannabes don’t know how important this benchmark is to actual customers.”
One of those thoughts is exactly your point. The other thought is that I disagree with the ARM vendors’ assessment of their customers — I think that the major customers will care deeply, (a) because performance on synthetic loads will help to decide whether it’s even worth thinking about testing performance on real loads; and (b) because all the X86 server vendors publish and tout their own benchmark performance, and will never let the missing ARM benchmarks go unnoticed by the potential customers. So if they are serious, then at some point, they will have to publish benchmark numbers.
Which led to my conclusion: either they are not serious, or they simply are not yet ready to show the world how serious they are.
@uma:
The more things change the more they stay the same. I actually had a Transputer board back in the day. I thought it was cool and was going to catch on. But it didn’t. Kind of like Modula-2.
Of course, Tilera is somewhat different than the Transputer — it has shared memory, as well as message passing. And Tilera comes around at a different point in the evolution of computers — many multi-core computers have already been produced, and many algorithms have been parallelized, and with things like video and Microsoft Kinect-style 3D motion recognition, ordinary consumers are doing computation these days that is highly parallelizable and can make use of multiple cores. And although Occam seems to have fallen by the wayside, Haskell, Erlang, Google’s Go, etc. inherit enough from it in spirit to help programmers actually use parallelism more easily.
So, it may be that the Transputer was just too early for its day — that, as an underfunded niche technology, it couldn’t keep up with the fast improvements in single-threaded performance, and the parallel mesh seed ideas just needed to wait and germinate at a point in time like now, when single-threaded performance increases have seemingly slowed.
OTOH, if you look at where X86 is now, it can already do NUMA or UMA, and other message passing, via HyperTransport. And as I mentioned before, X86-64 is a much more orthogonal architecture than earlier versions, and it wouldn’t surprise me if, at some point, X86-32 and X86-16 decode was dropped from the die and only supported via emulation. So to paraphrase the old saw “I don’t know what the future of networking will be, but I know it will be called Ethernet,” I will say “I don’t know what the future of CPUs will be, but I won’t be surprised if the mainstream high performance ones are called X86, or if they are made by Intel.”
Intel had a prototype 80-core CPU back in September 2006.
http://news.cnet.com/Intel-pledges-80-cores-in-five-years/2100-1006_3-6119618.html?tag=mncol;txt
Note that Intel promised production units “in five years”.
Still waiting.
I have a good friend who does all the linux (kernel) work for Calxeda. (I’m kinda proud that I gave him a previous job that turned him into a kernel engineer.) I’d say the jury is still out on non-x86 in the data center.
While Microsoft has promised Windows 7 on ARM, it’s also pressing Intel to build a 16 core Atom-based CPU
http://www.tomshardware.com/news/16-core-Atom-SoC-Dileep-Bhandarker-Xeon,12102.html
The future in servers still looks like x86.
@Patrick Maupin:
> And although Occam seems to have fallen by the wayside, Haskell, Erlang, Google’s Go, etc. inherit enough from it in spirit to help programmers actually use parallelism more easily.
Haskell and Erlang are very powerful to those who understand them. You have to have a pretty good background in math, lambda calculus, automata theory, parallel computing to understand how these these kind of languages work beneath the surface. Recalibrating the average programming programmer from his imperative C/C++/perl/python into the Haskell way of doing thing is not going to be a fun excercise – if at all achievable. How many programmers have you met who are lisp hackers or whose brains are wired the functional programming way? Not many. That was the very thing that made me turn away from writing code many years ago after doing a couple of years (tour of duty) in the software industry. My brain was wired the functional programming way.
The reason I advocated the D language earlier (and was corrected by others that Go was better) was out of pragmatism because clearly this parallel future is ahead of us. Ideologically, I am very much in the Haskell and Erlang camp. I think though that you’re too hopeful if you think the software world is all of a sudden going to have a change of heart and move to these languages even in era where we may well have 100 core CPUs in our laptops and desktops in the not so distant future. The software world has so much inertia.
> it wouldn’t surprise me if, at some point, X86-32 and X86-16 decode was dropped from the die and only supported via emulation
I have always wondered why they don’t that. Nothing can help intel more than doing this one thing.
>This is not a company with all their eggs in Google’s or anyone else’s basket.
Of course, that is one of the benefits of open source, you don’t have to rely on anyone else’s basket.
>It is long ago that I met people that actually liked to use Windows.
That depends on what you are comparing it to. Every OS (or other technology) has weaknesses and problems, a successful technology for a given purpose depends on fitting the technology to the user’s goals and abilities. I actually prefer Windows to Ubuntu–I dislike GUIs, but if I have to use one, Windows (even Vista) “just works” more reliably and predictably than Ubuntu. It didn’t help Ubuntu at all that it does not come with emacs or support dial-up internet. I’ll stick with Slackware and, for now, Vista. As I have written before, Vista has annoyed me enough that I am migrating gradually to just Linux, but with dial-up service that is hard.
@uma:
I don’t. Not in the slightest. I think we’re farther along than when I bought my Transputer, and that there are network effects. Parallelism helps with video stuff, better hardware helps with parallelism, better software makes it easier to control the better hardware, etc.
In short, Moore’s law feeding into this circle.
But…
I think that single-threaded performance is still extremely important. Because that’s how people think, as you point out. But also because there are a lot of tasks that don’t lead themselves to easy parallelization. Just as with human organizations, there can come a point where the cost of communication outweighs the benefits of multiple people trying to do the work.
And, I think X86 not only has the lead over all contenders there at the moment (for affordable general purpose processing) but is likely to maintain or even extend that lead over the medium term. One of the reasons I believe this is that I see some of X86’s supposed shortcomings as huge benefits.
For example, the limited register set of the original X86 was a “shortcoming” that supposedly couldn’t even stand the onslaught of the Motorola 68000. But software compilers got much better much more quickly at register-spill algorithms than they would have on a larger register set. And hardware implementations got good at register renaming. And the working set required to be saved on a task switch got smaller. All these things are performance enhancements that are, for the most part, independent of architecture, but evolved more quickly because of the architecture.
Branch prediction is another example. Useful and performance enhancing on an architecture that supports predicated instructions, but absolutely required for performance on an architecture that doesn’t.
So, essentially by simply picking an ISA and getting on with it, independent advances could be made in hardware performance and compiler performance. While this was going on, most of the other vendors were bragging about how they co-developed the hardware and software, and so the architecture was absolutely “perfect.” Unfortunately, perfect is often a codeword for unchangeable and brittle, and when things weren’t quite perfect, the vendor would change both the hardware and software together, requiring recompiles for the OS and every application. In other words, a network effect killer, while the X86 architecture was blessed with the dual benefits of network effects and robust competition.
Finally, there is one other architecture (ARM) with enough interest to engender similar network effects and implementation competition, but at the same time there is now enough open source software available to mean that recompiling for a new architecture is no big deal (so there could even be room for a few more CPU architectures), and there are enough high performance VMs available to mean that a lot of software doesn’t even need a recompile.
So, the next few years should be interesting, but rightly or wrongly, my gut feel is that the X86 vendors still have a huge lead on design, verification and validation of the sort of speculative, superscalar, out-of-order execution tricks that it takes to win on the performance front.
Erlang requires less of a background in math. The biggest toe stub features of Erlang are non-mutable memory, limited classes (and classes without mutable memory aren’t that interesting anyway). I’m hopeful that Rust will contain enough of the Erlang mindset to get the majority of benefit while being mostly compatible with imperative thought processes.
What i’m hoping to see from Rust is that if you’re creating a serial task it looks more or less normal, the changes occur when you start to incorporate parallelism.
Intel had a prototype 80-core CPU back in September 2006.
Apple’s been telling their developers to use Grand Central Dispatch and OpenCL to be prepared for such systems. When I talk to my old colleagues there, they’re talking about dozens of cores Real Soon Now.
@Some Guy:
> When I talk to my old colleagues there, they’re talking about dozens of cores Real Soon Now.
Given that you could buy a dozen cores over a year ago if you had crazy money, and that you can now buy half a dozen cores now for cheap, I have no problem believing this for some reasonable value of RSN.
The go language does have object-oriented features. I wouldn’t call it a prototype-based language like javascript or self — you don’t take an existing object and modify it to get new ones. Instead it’s just “define methods”. What go doesn’t have is an object hierarchy. You can’t “inherit”, though you can of course compose by having another object as data. Interfaces are just a list of methods that must be implemented. If they’re implemented them, so is the interface, without any need to declare it. If you want to force declaration of implementing an interface, the workaround is a do-nothing method for “tagging”.
About the SPEC benchmarks: how do the various JVMs needed to run ssj_ops cope with the ARM framework? Wikipedia tells me that the Sun/Oracle Hotsport port to the ARM is a “zero-assembler Hotspot”, which maybe means that it runs like a hot dog.
Possibly the reason the ARM vendors don’t submit to SPEC tests is that they don’t want to promote absurdly prejudicial comparisons between ARM and x86.
No. x86 is the worst of all possible worlds — Gabriel’s “Worse is Better” applied to CPU architectures. It won because of IBM-Microsoft hegemony and not for any other reason. The register-starved nature of x86 actually set back compiler technology quite a bit, as no one was willing to work on anything besides C compilers because that’s what the vast bulk of the code was written in. Compilers for other, more interesting languages (*cough*Lisp*cough*) simply didn’t get worked on nearly as much because it was too much time and effort. That’s changing now — technologies like LLVM helped make weirdly broken CPU architectures more tractable I think — but the RISC winter had a chilling effect (heh) on programming language development.
The dominance of x86 may well have contributed to the dominance of C++. Do you want to cite that as a boon to mankind? :)
The x86 and 68k are the Bush and Gore of CPUs: the other guy should have won.
Funny you should talk about “the other vendors” when the biggest boondoggle of this type was committed by……. Intel, with the Itanium.
Not much choice here in corporateland … as you pointed out, the suits live in Outlook.
>. It won because of IBM-Microsoft hegemony and not for any other reason.
Actually, Intel’s ability to improve the underlying process for fabricating the x86 had a lot to do with masking its architectural deficiencies. For quite a long time there, they were able to just keep jacking up the clock rate.
Jeff, you should also keep in mind that GCC did a lot to retard compiler advancement. Back in the late ’80s, there was a robust, competitive market for compilers. I remember Aztec C, Lightspeed C, Lattice, Borland, and many others. Once GCC came on the scene, those guys all went out of business, and compiler development dwindled to the CPU vendors working on their optimizers, and few academic projects like PCCTS.
Luckily, Chris Lattner had the right academic project (LLVM) to offer to Apple when they were shopping for a way out of the dark ages that GCC inflicted on us. They hired him and reassigned the people that they’d had working on GCC, and expanded the team out to something over 30 full-time developers, so today we have a beautiful modular compiler infrastructure that’s making all kinds of things possible.
>The x86 and 68k are the Bush and Gore of CPUs
Oh, come on! The 68K had its faults, but that’s way too harsh.
>the quants at the big banking houses and AIG were mathematical geniuses, weren’ they?
They are, but the trap they fall into is what Mises called the Knowledge Problem when he described the impossibility of central planning producing better results than a free market. To predict the market, you have to predict human behavior, and you simply can’t gather enough knowledge to do that. What they end up with might as well be astrology, since math has to be based on valid premises to be valid.
@Jeff Read:
In addition to all the stuff Some Guy said, I have to take particular exception to this. C++ came from C which came from the Unix world. I was working on PCs back in the day, and it was assembler and Pascal and some FORTRAN, then Modula-2. C killed Modula-2, and that’s what let C++ live. But the only thing that gave C a foothold was its deep Unix/University roots.
In short, C and the X86 grew up in completely different worlds, and only got married later. In fact, if you look at some of the more ambitious X86 instructions like ENTER and LEAVE, they were specifically designed to support nested scopes on the stack. No, C and RISC were made for each other.
BTW, I have nothing against ANSI C as a high level assembler. Like any good natural language, it borrowed some of the best bits from other languages, and morphed into something quite usable.
esr:
>I think the least hypothesis for why there aren’t SPEC ARM benchmarks is that the ARM vendors don’t think their potential customers >care about SPEC benchmarks.
The early adopters for this, the guys with huge homogeneous datacenter farms, probably don’t. A lot of the second wave would, though; we use it a good bit at (the large financial services firm where I) work despite its limitations because it seems to be the best ‘generic’ CPU benchmark available. An eye-popping SPEC score wouldn’t be strictly necessary, but it would certainly be a big plus if you were trying to get a proof-of-concept test in here.
You can get a copy of the benchmark suite for $500 – a non-vendor could run it themselves, if they care.
There are no recent results for anything but x86, Power, and Sparc.
Patrick Maupin Says:
> FORTRAN, then Modula-2. C killed Modula-2, and that’s what let C++ live. But the only thing that gave C a foothold was its deep Unix/University roots.
I remember the history to be much simpler than this. C++ stole the market because it was adopted by the monopoly money machine MS. May as well say its victory was dictated from above. In any event, pity something sane like Modula-2 couldn’t have won, we’d all be much better off now.
@Michael Hipp:
> C++ stole the market because it was adopted by the monopoly money machine MS
Sure, but the network effects went both ways — if MS hadn’t decided that C was the right thing to code Windows in (based partly on what programmers were learning in college, and probably partly on the fact that it got a good deal on the Lattice C compiler) then C++ wouldn’t have been obvious as a successor.
>Sure, but the network effects went both ways — if MS hadn’t decided that C was the right thing to code Windows in (based partly on what programmers were learning in college, and probably partly on the fact that it got a good deal on the Lattice C compiler) then C++ wouldn’t have been obvious as a successor.
I think the reasons for C’s dominance are deeper and less contingent than anyone here is talking about.
There’s a marvelous book called Computer Architecture: Concepts and Evolution by Gertrit Blauww and Fred Brooks that traces the historical evolution of instruction set architectures. One of the book’s principal themes is that there is a “standard architecture” which is a strong attractor in ISA design space that keeps getting rediscovered in successive technology generations. C is, despite various imperfections, an extremely good fit for low-level programming on the “standard architecture” while being just abstracted enough so code can readily be moved among variants of it. It thus occupies a sweet spot in language design space that is implied by the attractive power of the standard architecture.
this might be interesting:
http://www.bgr.com/2011/05/09/previous-generation-iphone-3gs-ipad-still-outsell-new-android-devices-analyst-claims/
“Walkley writes that Apple’s iPhone 4 is far and away the top-selling smartphone for both AT&T and Verizon Wireless”
This contradicts the Thunderbolt outselling the iPhone 4 on Verizon….. as did the Verizon earnings release.
@phil:
> This contradicts the Thunderbolt outselling the iPhone 4 on Verizon
Perhaps not. The Thunderbolt survey was apparently done in a short time period:
http://vritesh.com/smackdown/htc-thunderbolt-challenging-iphone-at-verizon.html
> Walkley writes that Apple’s iPhone 4 is far and away the top-selling smartphone for both AT&T and Verizon Wireless
That article meshes somewhat with other data we have seen, like the good technology report that showed that, even in March, older iPhones on AT&T were kicking Verizon’s iPhone 4 butt.
In any case, Apple’s handset models have to be selling comparatively well for those two carriers. They only sell a couple of models, and total Apple sales in the US last quarter weren’t too much less than combined sales of Samsung and HTC, who both sell to other networks.
“Perhaps not. The Thunderbolt survey was apparently done in a short time period:”
There was a lot of talk on here about how the Thunderbolt as outselling the iPhone 4 with no qualifiers. But yes, the Thunderbolt survey was for a 2 week period as I recall – and only Verizon stores.
@esr
> C is, despite various imperfections, an extremely good fit for low-level programming on the “standard architecture” while being just abstracted enough so code can readily be moved among variants of it. It thus occupies a sweet spot in language design space that is implied by the attractive power of the standard architecture.
Absolutely. And it is likely C will be with us for a long time yet for that reason. C is very well designed for its intended usage: as a portable replacement for assembler.
What puzzles me is why no such “sweet spot” has emerged at the next level up and been filled by a similarly well-designed language. I keep thinking Google’s Go is the best contender we’ve seen in 2 decades. But kicking C++ out of that spot seems even more difficult than dislodging Windows and MS Office from the desktop. At what point do network effects just become monopoly totalitarianism?
@esr:
Some of them. But AT&T’s partnership with academia was actually one of the prototypical open source developments, and led to essentially free operating systems and compilers when other vendors were still charging exorbitant sums.
No doubt. But so (IMO) was Modula-2. The thing is that ANSI C sucked in most of the “must-have” goodness from other languages while still allowing backward compatibility to old-style K&R C. It allowed somewhat better programming practices without requiring them. And what goodness ANSI C itself didn’t suck in, the compilers and linters mostly took care of, with features like warnings about things like “if (x=1) foo();”.
So C itself (not C++) is definitely an evolutionary good-enough success story. As is X86. Personally, I don’t think that backward compatibilty hurts X86 on power all that much, but if, as previously discussed, Intel or AMD finally jettisons some of the past and builds a 64 bit only X86 processor (which is fairly viable now that there are open source BIOS implementations and 64 bit operating systems), I think we will quickly discover that the primary power hogs in the X86 space are those that would be power hogs in any performant architecture — cache, speculative execution, fast front-side busses, etc.
I am to context-switch back to RIM. The blackberry playbook apparently does not even have an e-mail client. They might finish their work on the e-mail client before year’s end.
Some commentator pontificated: “A blackberry without e-mail is like a wedding without the bride”.
That is how much trouble RIM is in.
Guess this is old news. looks esr had blogged about this earlier. my bad.
The blackberry playbook apparently does not even have an e-mail client.
Holy crap, I nearly fell out of my chair when I read that. These are the guys who are running ads about how their product is “professional”?
Don’t count Microsoft out of mobile yet. With their Skype acquisition, they are poised to become the major player in VoIP, which they can leverage directly into a mobile strategy, providing cheap voice calls over Windows Phones (and possibly iPhones; I expect the Linux and Android versions of Skype to go away).
Jeff,
Looking at the history of Microsoft’s acquisitions, I would say that today’s news doesn’t make MS a major player in VOIP, so much as destroy everything Skype has achieved to date. I can see the way this will play out: some idiot in MS senior management will decide that migrating Skype’s server infrastructure to Windows Server is absolutely necessary, and in the process of this migration, Skype will have extended periods of downtime.
Their engineering talent will melt away as people stream out of Skype to go work for Cisco, Google, Apple, anywhere else they can go. Meanwhile, the client apps will get bloated up or crammed in with MSN Messenger, and you’ll have to install Silverlight to install Skype, etc, etc.
If there’s any hope for Skype to survive, it will be with them getting spun back out in a year or two like Bungie did.
>Looking at the history of Microsoft’s acquisitions, I would say that today’s news doesn’t make MS a major player in VOIP, so much as destroy everything Skype has achieved to date.
That was my reaction too. Microsoft’s record at such technology acquisitions is very poor. Or should I say “Danger, Will Robinson! Danger! Danger!”
There already is a third architecture — Power. It’s currently not a player in the desktop/tablet/phone area, but given its range (everything from mainframes to gaming consoles to embedded controllers), the significant numbers it sells, it’s not going away, either.
Is the reality distortion field this powerful, or does Apple simply pay “journalists” more than their supposed employers?
http://www.zdnet.com/blog/btl/on-fragmentation-google-still-comes-up-short-against-apple/48569
Android’s app store is expected to overtake Apple’s by August, companies who care have no problem maintaining both Android and Apple versions, most of the devs probably cut their teeth doing desktop or web apps running on wildly divergent platforms, and the best response they can come up with is that fragmentation is Android’s Achilles heel?
In other news, I think I just figured out where some of the stronger than expected Apple performance against Android in Q1 of this year (that we see in some data and not others) came from.
This article claims that the smartphone market actually contracted after Christmas, but that Apple managed to forge ahead with huge gains anyway. Their numbers don’t exactly match some other numbers, which showed market growth; nonetheless I think they’re on to something. They actually got very close to figuring it out:
But then they dropped the ball on further analysis. Occam’s razor says that Apple, with its own stores and its tight carrier relationships, has much tighter inventory control than most of its competition. (Compare and contrast Samsung’s shipments to dealers of its tablet, which by most accounts was a debacle on “sell-through”.)
The Android vendors were cranking out and shipping all they could before Christmas, whereas Apple probably keeps less inventory in the pipe, so any first quarter “adjustment” would primarily have been to dealer’s stock of Android (or at least non-Apple) phones.
@Steven Ehrbar:
> There already is a third architecture — Power.
Power is not bad, and IBM is actually pushing it fairly hard — even making it available for free under an academic license.
There are actually a few semi-viable 32 bit architectures. Besides Power, there’s MIPS and SPARC (and its Leon implementation, as well). There are even a couple of completely open source ones. Don’t know if they will become competitive, but there is a lot of preexisting work that can be leveraged that could make it possible.
I could be wrong, but I think that, just as Android is kicking ass because it’s royalty-free, any architecture that has a snowball’s chance of displacing ARM in the phone space is probably also going to be royalty-free, by being open source, or licensable for a reasonable fixed sum, or developed internally by one of the silicon vendors.
@Some Guy
People might end up possibly end up leaving skype and moving over either to yahoo messenger or google voice and pushing for skype-like features in google voice. Also, from what I understand yahoo has access to the skype codecs which made skype popular to begin with and made it the golden standard in VoIP voice quality. Perhaps an exodus to Yahoo Messenger too.
Have you guys been reading about the Android Open Accessory Toolkit?
Many, if not most, Android devices are USB slaves (“devices” in the vernacular), not capable of on-the-go or host operation.
Google solves this by creating a specification (and an underwhelmingly powered reference design using the Arduino, to drive the point home that not much hardware is required) for a fairly dumb “host” “master” that actually takes its direction from the Android “slave” device.
It seems pretty cool. If I ever get around to doing something like this, and it’s not so simple that I just use an 8 bit controller, I’d probably go with something like an LPC3131 running nuttx.
Good article about the meaning of Google’s recent moves:
Some Guy, you are trolling. And you are not being amusing. Stop now.
Oh, please. And you accuse me of fanboyism? Desktop linux failed: grow up and deal with it.
>Oh, please. And you accuse me of fanboyism? Desktop linux failed: grow up and deal with it.
You are capable of being interesting, but my tolerance for contentless baiting is limited. This was contentless baiting, and if you persist in it I will ban you.
>my tolerance for contentless baiting is limited.
Fine, eric. Have your little echo chamber where linux didn’t fail at “world domination”.
Some Guy,
The available evidence does not support your hypothesis.
Well, later Patrick said this:
> I think we both agree that exclusively selling retail PCs with Linux instead of Windows is probably suicidal, except in special cases for special businesses. And as long as Microsoft, with their “marketing dollars”, can punish vendors who dare to sell Linux in addition to Windows, we will never know what the marketplace would think of stores that sell both.
Which is a pretty good clue that Patrick thinks Linux did fail at “world domination”.
esr did not accuse Patrick of contentless baiting.
So it seems there is something more subtle than a desire for an echo chamber.
Yours,
Tom
Scooping the Wharton School:
http://knowledge.wharton.upenn.edu/article.cfm?articleid=2793
Buffet himself has said this… when asked once what his secret was, he replied “I always sold too early”… which I took to imply the corollary that he never sold too late. Choosing to give up the last little bit of potential profit for a sure profit now.
–Shannon