The Smartphone Wars: China syndrome strikes!

A few weeks go, I reported a conversation with a Chinese-American investor who confirmed my suspicion that the next game-changer for the smartphone market is going to come out of mainland China.

He wouldn’t name the company, but I think they’ve just unstealthed. ICube matches the hints he was dropping exactly.

And what is ICube touting? Unit-cost and power-efficiency advantages from an Android SoC. It’s exactly the scenario I’ve been predicting. No volume production planned until 2012, but it means the SoCs that are supposed to be coming out in 3Q2011 from Broadcomm and Nvidia will, not long after launch, have price competition from an outfit that can tolerate lower margins.

And that makes three. Not that these will be the last chips in their class. Heck no – color me astonished if we don’t see two or three very similar announcements in the next six months, some from startups and some from players at the scale of ZTE and Huawei and Samsung. Potential unit volume is, after all, in the billions. That’s billions, not milllions.

120 thoughts on “The Smartphone Wars: China syndrome strikes!

  1. Absolutely agree on that market of billions thing.

    The low cost smartphone, will be the pervasive computing platform of Asia.

    When India and China, which already have pervasive mobile networkis in major population centers; get low cost networked computing and communication devices on a large scale…

    That’s a world changing thing.

  2. In many third world countries, the capital does not exist to lay landline broadband like cable, fiber, or DSL over copper. But building cell towers is within their means, so wireless leapfrogs over POTS. We’ll probably see the analogous leapfrog of smartphones over desktop computing.

  3. It appears we can take these cheap-but-capable Android phones as a gimme during the next 12-24 months. Yay.

    So when are we going to get some cheap-but-capable data plans to go with them?

    Even at today’s hardware prices, that’s the only thing stopping me from converting my entire household to Android.

  4. @esr:

    Do have more info?
    They have decided to go with their own instruction set. That is bold. Are they bankrolled by the chinese govt at all ?

  5. > So when are we going to get some cheap-but-capable data plans to go with them?

    I think Virgin Mobile’s $25/month for unlimited data (that is, 5GB at full-bandwidth, then it throttles down if you go over), unlimited text messages, and 300 minutes a month qualifies as reasonable. There are also some offerings from Boost Mobile, Cricket, and MetroPCS that don’t look too bad.

    If you don’t live in a Sprint coverage area, or you demand to go with one of the Big Boys (Verizon, AT&T) then don’t hold your breath waiting for cheap anything.

  6. >Do have more info?

    No. You now know everything I do except the name of the angel investor who tipped me off in New York.

    >They have decided to go with their own instruction set. That is bold.

    I don’t actually think so. As long as they can port Dalvik, who cares what the instruction set is?

    >Are they bankrolled by the chinese govt at all ?

    Dunno. It’s not unlikely.

  7. I don’t actually think so. As long as they can port Dalvik, who cares what the instruction set is?

    This i think is the biggest thing that Android brings.

    I don’t care what you’re running on, I don’t care what your OS is, I don’t care about anything other than if you make yourself compatible you’re a functional part of my ecosystem.

    In before the comment about Android Fragmentation.

  8. >I don’t actually think so. As long as they can port Dalvik, who cares what the instruction set is?

    Game developers do.
    Given how crappy gcc-generated arm (neon) code tends to be, I can routinely get 300-600% speedup just by rewriting certain core functions in assembler.

  9. Very cheap SoC’s with android? That means a large bunch of appliances running Android that:

    1 Will connect over WiFi. Expect free WiFi/WiMax comming everywhere where they want you to be around. That is, restaurants, pubs, coffee shops, public transport (we already have free WiFi in long distance trains), libraries, malls, shops, etc.

    2 That will only require your SIM card to connect over the air. Children around already swap SIM cards to make/recieve calls and text messages (“can I borrow your phone for a minute, my battery is dead”). So what easier than to stick your SIM card in your laptop/tablet to get Internet?

  10. If you don’t live in a Sprint coverage area, or you demand to go with one of the Big Boys (Verizon, AT&T) then don’t hold your breath waiting for cheap anything.

    Read: don’t live in a major city or along an Interstate.

    /me raises his hand.

    Sprint’s about to have major problems if the AT&T/T-Mobile buyout happens.

    As long as they can port Dalvik, who cares what the instruction set is?

    And as long as Oracle doesn’t succeed in squashing Dalvik…and as long as there’s a way for people to do native code builds for stuff for which Java is poorly suited…

  11. > So when are we going to get some cheap-but-capable data plans to go with them?

    As the target market for these cheap SoC Android phones seems to be Asia, I cannot see why the US network pricing plan is relevant?

  12. “Apple created Android, or at least created the conditions necessary for Android to come into being” –Stephen Elop

  13. >As the target market for these cheap SoC Android phones seems to be Asia, I cannot see why the US network pricing plan is relevant?

    At least as I see it, the expectation is that, any capability that’s within the reach of poor Asian peasants should also be within the reach of Western suburbanites. Why should the peasants get all the goodies? (Asia may be the target market, but it’s a global marketplace you know.)

  14. > Why should the peasants get all the goodies?

    Asian city dwellers might get more cellphone bang for their buck than those in the USA. So, Indians and Chinese might get better data plans than those in the USA.

  15. >65nm? HAHAHAHA! It’s going to be a pig.

    Perhaps, but a pig with a high yield rate. Thus a very inexpensive pig, even from first silicon. This makes sense under the assumption that they’re aiming at cheap Androids for the Third World and BRICs. They don’t need to set any speed records to address the low-end market effectively, and the price pressure will act on fast Androids just the same.

  16. What percentage of the Android market as a whole do you think iCube needs to capture in order to be cost-effective?

  17. @not(Andy Rubin):

    > 65nm? HAHAHAHA! It’s going to be a pig.

    Depends on how much they’ve integrated. If they use external flash and RAM, could still be fine. Also, if they’ve integrated analog, that gets more difficult the smaller you go.

    I work for a profitable fabless semiconductor company, and the smallest we go is 130 nm.

    @Bryant:

    > What percentage of the Android market as a whole do you think iCube needs to capture in order to be cost-effective?

    Depends on how much advantage they get by being a fairly early mover in the lower-cost space. But there were 37 million Android phones shipped last quarter. I think if they can start out at 2 million units/quarter with decent pricing and margin, they can grow like gangbusters. But…

    They have to be prepared to put 90% of the profit back in the business. As competition heats up, not(Andy Rubin)’s point may actually become relevant. In addition to innovating on features, they will have to spend their lives on the cost-reduction treadmill, just like the rest of us silicon types.

  18. (Background for the uninitiated)

    65 nm and 130 nm are process nodes. The number is a distance that roughly defines the size of the smallest feature on the chip. Since chips are two-dimensional (for now — see where Intel is going with cache) the number of transistors you can cram into an area on the chip is roughly equivalent to the inverse of the square of the feature size, so you would expect that a piece of silicon at 65 nm would hold roughly 4 times as many transistors as the same sized piece of silicon at 130 nm.

    But there are a lot of other tradeoffs, as well. Mask sets, which are fancy lenses with lines on them used for exposing the wafers in the various processing stages, cost exponentially more the lower the geometry. You can spend millions on masks before you even try to start building a chip. Or tens of thousands. It really depends on the geometry.

    Then defect density, which is one of the factors affecting yield, will go up with smaller geometries as well, but it may not go up as fast as the feature size shrinks. That partly depends on the fab, the process, the phase of the moon, etc. Poor yield can be a major factor contributing to cost. Especially if you package your parts before you test them, because the package often costs more than the silicon.

    There are also power issues. Typically, when you move to a smaller geometry, dynamic power (when clocking) goes down, but static power (drawn all the time) may go up because the smaller transistors get leaky. (This is why we had the discussion a few weeks ago about FinFETs and Intel’s 3D transistors.)

  19. > I work for a profitable fabless semiconductor company, and the smallest we go is 130 nm.

    Please tell me about all the cell phone SoC design wins you have. You and I both know that they can’t run
    the clock period down without the leakage current going sky high.

    2007 called, they want their feature size back.

    The whole things smacks of “Loongson makes a smartphone SoC” to me. It’s all there. The heavy MIPS involvement, the “we’re doing our own ISA” (to avoid the MIPS patents, and thus the need to pay MIPS for a license”, and the recent announcements from Action Semi and Ingenic Semi (oh, hey, more companies supported by the Chinese government) about their respective 1.3GHz and 1.0GHz Android SoCs.

    http://www.nasdaq.com/aspx/company-news-story.aspx?storyid=201105310900PRIMZONEUSPRX___223192
    http://www.nasdaq.com/aspx/company-news-story.aspx?storyid=201105310900PRIMZONEUSPRX___223193

    Could be iCube doesn’t want to be bound by the Google anti-fragementation agreements that all processor vendors MUST sign, too.

  20. >What percentage of the Android market as a whole do you think iCube needs to capture in order to be cost-effective?

    No idea. I’ll bet there’s a pretty firm estimate in ICube’s business plan, though.

  21. >In addition to innovating on features, they will have to spend their lives on the cost-reduction treadmill, just like the rest of us silicon types.

    Yes, I took that as a given. I think their plan must be to come fast out of the gate and use the early revenue to downscale the design to a smaller lambda. It’s exactly what I’d do in their shoes.

  22. > Mask sets, which are fancy lenses with lines on them used for exposing the wafers in the various processing stages, cost exponentially more the lower the geometry. You can spend millions on masks before you even try to start building a chip. Or tens of thousands. It really depends on the geometry.

    I imagine that the used (or cloned) Applied Materials gear is cheap.

    But they can still use little aluminum legs in the interconnect. (WHADAJOKE!)

  23. I’m sure there will be plenty of Android SOCs coming, but that’s burying the lede on iCube.

    If you read closely, this is not a standard phone chipset integrated onto one wafer – that is, rather than a CPU and GPU on the same chip it seems to be some kind of multicore chip with cores that can perform either CPU or GPU functions. This would explain the nonstandard ISA (not the need to avoid MIPS patents).

    So, it’s the return of software rendering – that strikes me as a bad idea, offhand. On the other hand, the ability to rebalance load between GPU and CPU may be a big win, and maybe software rendering will be OK on a cellphone-size screen with an ISA optimized for it.

  24. “Apple created Android, or at least created the conditions necessary for Android to come into being” –Stephen Elop

    Great, so now Elop confirms not merely that he has no plan forward but does not understand the past history of his new industry either.

  25. > use the early revenue to downscale the design to a smaller lambda.

    perhaps Patrick will explain why his company hasn’t done this.

  26. I don’t actually think so. As long as they can port Dalvik, who cares what the instruction set is?

    Android apps with any kind of performance requirements (including games, audio processing apps such as the “T-Pain” vocoders, etc.) have significant ARM native-code parts.

    Oracle is seeking enough damages in its copyright and patent suits to make Android an overall significant net loss for Google. It’s quite possible that Dalvik will have to be abandoned altogether in order to make Android non-infringing.

  27. use the early revenue to downscale the design to a smaller lambda.

    perhaps Patrick will explain why his company hasn’t done this.

    I think it’s pretty obvious from some of the stuff I wrote that it’s not always necessary or desirable. If the package and testing costs swamp the silicon costs and you’re in a fairly stable market, the math on the ROI of going to a smaller node is pretty easy.

    Those guys aren’t going into a very stable market, though…

  28. @spqr:

    Elop may overstate his case a bit, but I think it is safe to say that Apple’s actions helped create market conditions that hastened Android’s adoption greatly. There is no question that Apple disrupted the market, or (and this might be what Elop is missing) that Android was already in development and basically present and ready to participate in the disruption.

    The question is whether Android could have caused the disruption itself (without Apple) in the same sort of timeframe. And it’s easy to count arguments why google on its own might not have pulled this off anywhere near as quickly.

  29. @Jeff Read:

    The Judge Alsup already reduced the number of Oracle’s claims from 132 to 3 and Google’s prior art claims from 200 or so to 8.

    It’s looking like Google is in this case to fight. They aren’t going to settle. And Oracle’s remaining 3 claims are all that are going to trial. It’s not looking good for Oracle.

  30. Something I’ve always wondered: How easy is it to take an existing design, and “just” shrink the process and “just” run it faster? Is it “just” a matter of shrinking the die down, or do you have to make pervasive, systematic changes across the entire chip, or something in between, or what?

    (It doesn’t matter much if they start shipping at 65nm if it’s the technological moral equivalent of flipping a switch to shrink it down in three months when it is profitable to do so.)

  31. Patrick — thanks, and also thanks for the informative discussion of process nodes. (Sincerely. I wish you had your own blog; I think it’d be a really interesting place for productive conversations.)

    If 2 million units a quarter is enough to get competitive pricing, shouldn’t Apple be able to do the same thing with their designs if they want to? They’re pushing way more than that. “If they want to” is not a hypothetical, of course: it may be the case that Apple values speed or power consumption over price.

    I will also note that while I agree SoCs are a big deal, I am officially reluctant to believe in timelines until I see chips in production. E.g., the last time Eric got excited about a chip announcement it was from Broadcom and phones based on it were supposed to be shipping by now. Apple may have hit delays due to Qualcomm, but it’s not like the rest of the world is immune to that kind of thing.

  32. @Cathy: “If you don’t live in a Sprint coverage area, or you demand to go with one of the Big Boys (Verizon, AT&T) then don’t hold your breath waiting for cheap anything.”

    @Jay Maynard: “Read: don’t live in a major city or along an Interstate. /me raises his hand.”

    What fraction of the US population (not fraction of land area) is covered by one those criteria? (Hint: large) And it isn’t necessary to live in a major city as long as you live in the metro area of a major city. That’s a major difference.

    I agree that it’s tough on you as an individual, but in terms of reaching a sizable audience it can be quite effective without incurring the high costs of serving low population-density areas. Rural phone service in the U.S. was always subsidized by urban phone service during the period of Ma Bell monopoly regulation, and I don’t think it would have been universal without that regulation.

    @Jay: “Sprint’s about to have major problems if the AT&T/T-Mobile buyout happens.”

    Sprint senior leadership has already said as much. If AT&T is allowed to buy T-Mobile, we will probably rapidly move to a duopoly and service will become expensive and poor. I really hope that doesn’t happen, as I like my new smartphone but would reluctantly give it up completely if my only choices were AT&T or Verizon at current data plan prices.

  33. >E.g., the last time Eric got excited about a chip announcement it was from Broadcom and phones based on it were supposed to be shipping by now.

    Er, no. I wasn’t expecting SoC phones in volume until third quarter.

  34. >It’s not looking good for Oracle.

    No, it isn’t. Apparently this case got a judge with a clue. It does happen.

  35. >Something I’ve always wondered: How easy is it to take an existing design, and “just” shrink the process and “just” run it faster? Is it “just” a matter of shrinking the die down, or do you have to make pervasive, systematic changes across the entire chip, or something in between, or what?

    Something in between, but usually closer to the “just shrink the die down” end. After you shrink the design, you have to test for bugs introduced by inductive coupling and other modes of crosstalk. Sometimes parts of the design have to be reworked, which is expensive. Depending on the fab, it might be that you lose so much yield that the downscaled design isn’t viable.

    Patrick Maupin may have more to say about this. My knowledge is all secondhand; he’s been up to his eyeballs in the actual engineering.

  36. @bryant:

    Patrick — thanks, and also thanks for the informative discussion of process nodes. (Sincerely. I wish you had your own blog; I think it’d be a really interesting place for productive conversations.)

    Thanks, but I’m far too lazy for that. Maybe in my next life.

    If 2 million units a quarter is enough to get competitive pricing, shouldn’t Apple be able to do the same thing with their designs if they want to? They’re pushing way more than that. “If they want to” is not a hypothetical, of course: it may be the case that Apple values speed or power consumption over price.

    There is no question that Apple pulls the volumes to get most favored nation pricing from all its suppliers. Apple is now the largest semiconductor purchaser in the world. There is also no question that Apple has been hiring chip designers and trying to cook something up.

    But there are potentially several reasons why the economics that apply to other companies don’t apply to Apple. For a start, Apple is in the US and pays a lot of its employees extremely well. Apple needs higher margin than a lot of other companies for its business model. Also, Apple’s marketing model demands immediate shipment of millions of units within the first week of shipping the first unit. That’s an excellent way to help get the higher margins, but it incurs some risks. A lot of companies like to ship smaller volumes for at least a few months to make sure there aren’t problems in the design or in the testing of the chips. A small Chinese company selling a few million chips to other Chinese companies will be in a good position to ramp volumes if everything goes well. Note that all these companies experiment with product mix and, unlike Apple, will have several designs in production simultaneously. This decreases margin but it also decreases risk, and they live in a cutthroat world.

    I will also note that while I agree SoCs are a big deal, I am officially reluctant to believe in timelines until I see chips in production.

    Couldn’t agree more.

  37. @Jeremy Bowers:

    Something I’ve always wondered: How easy is it to take an existing design, and “just” shrink the process and “just” run it faster?

    Sometimes you can do an “optical shrink”, e.g. from 180 nm to 130 nm. But that’s usually done for cost (size) reasons rather than speed.

    Usually, if you’re shrinking very much, you need a redesign for several reasons. The voltage will be lower, the oxides will be thinner, the optics get really squirrelly because you’re so close to the wavelengths of light that the shapes you use in the mask aren’t really the same as the shapes of the devices you’re producing, etc.

  38. @bryant:

    Just to make it clear, I’m not saying that Apple won’t be able to also benefit from doing its own silicon. Just that its business model is different enough that how it benefits vs. how Android handset manufacturers and Android silicon vendors can benefit is like comparing Apples and Androids.

  39. @Fackbook is evil:

    Shrinking below 65 nm is exponentially hard.

    Even 65 nm is hard. It’s incredibly disingenuous of not(Andy Rubin) to imply that no useful silicon is produced at that node.

  40. For largely digital ICs:

    Things can be shrunk much faster from node to node. Still though the change will be pervasive. Everything from timing, to clock-data recovery circuits, to clock trees and timing have to be changed. What makes things much more systematic and quicker is the EDA tools.

    For analog/mixed-signal ICs:

    It is a painstaking process. It takes a long time (years) to get the transistor models correct, and all effects that impact yields accounting for. There is generally a big mismatch between how fast digital can be made to shrink (from a design cycle perspective) and analog. Also analog, does not “shrink” with feature size as much as digital does. In other words, if your digital shrinks 40% (silicon area) analog does not tend to shrink that much. Analog also often includes cost adders (additional masks/processing) to make it yield. Unlike digital EDA tools, analog EDA tools aren’t as productive in terms of design cycle time.

    That is pretty much the summary.

  41. @Jay: “Sprint’s about to have major problems if the AT&T/T-Mobile buyout happens.”

    Sprint senior leadership has already said as much. If AT&T is allowed to buy T-Mobile, we will probably rapidly move to a duopoly and service will become expensive and poor. I really hope that doesn’t happen, as I like my new smartphone but would reluctantly give it up completely if my only choices were AT&T or Verizon at current data plan prices.

    As it happens, I was exposed to both sides of this argument recently, as someone – I don’t know who – felt compelled to focus-group this issue here in rural Minnesota.

    Sprint lost the argument with me, for a couple of reasons:

    1) Why is Sprint carping instead of competing? If they provided equivalent service to AT&T at a better price, I’d switch.

    2) AT&T’s 4G buildout map without the T-Mobile acquisition is a lot sparser than the map with it. In particular, it covers substantially all of Minnesota with the acquisition, but leaves out huge chunks of rural Minnesota, especially to the south and west, without it. For me, that’s a no-brainer.

  42. The distinction that uma draws between analog and digital is very real.

    digital really is analog. But, since there are just two signal levels, there’s a lot more slop allowed, and since APR (automated place and route) tools are used on the digital, you can consider that analog is more like assembly language, and digital is more like using a compiler (the tools are doing a lot behind your back).

    So, when it’s just a matter of changing the tool (and you get a whole new layout that you’ll probably be happy with), the design goes a lot faster than with analog.

  43. Pingback: Game Changer: Android ‘System on a Chip’ | TekNerve!

  44. Er, no. I wasn’t expecting SoC phones in volume until third quarter.

    Here’s the post. Date: December 27th, 2010. What you said: “Phones based on the chip are expected to ship in volume before mid-year 2011, with the first possibly hitting the trade show circuit in a month.”

  45. >What you said: “Phones based on the chip are expected to ship in volume before mid-year 2011, with the first possibly hitting the trade show circuit in a month.”

    Right, I see how you got confused; I was reporting Broadcomm’s projection, not mine, and wasn’t clear enough about the difference. I later revised my forecast to 3Q2011 due in part to issues with ramp-up of other components, notably including display glass.

    I think it’s actually pretty likely that Broadcomm met their target, based on reports of sub-$100 phones I’ve been seeing from overseas.

  46. Ah, fair enough. I stand by my note, mind you — the tendency is to slip ship dates throughout the industry. Or perhaps it’s just that marketing fluff is not reliable. Mmm, press releases.

  47. Thank you for the answers to my question. I’ve never been in an environment where I could ask that of people I trusted to give me an actual answer from experience before. :) I can talk out my own ass as well as any other not-an-EE.

  48. @Jay: “Sprint’s about to have major problems if the AT&T/T-Mobile buyout happens.”

    @Cathy: “Sprint senior leadership has already said as much. If AT&T is allowed to buy T-Mobile, we will probably rapidly move to a duopoly and service will become expensive and poor. I really hope that doesn’t happen, as I like my new smartphone but would reluctantly give it up completely if my only choices were AT&T or Verizon at current data plan prices.”

    @Jay: “Sprint lost the argument with me…why is Sprint carping instead of competing? If they provided equivalent service to AT&T at a better price, I’d switch.”

    Jay, you’re a sensible guy; you know that you can’t get something for nothing. Sprint *is* competing; what they’re doing is offering lesser service at a lesser price. For many of us, that’s a perfectly respectable trade-off. But they’re worried about being locked out by a Microsoft-style monster pair.

    “That alleged duopoly (Verizon and AT&T) would control 82% of post-paid subscribers, Sprint warns, more than 78% of wireless revenue, and 88% of all profits. ‘The Twin Bells’ market dominance would dwarf Sprint, the sole remaining national carrier, and the rest of the wireless industry, thereby creating an entrenched, anti-competitive duopoly,’ the filing contends.”

    http://arstechnica.com/tech-policy/news/2011/05/attt-mobile-merger-on-the-hot-seat-as-sprint-files-to-block-it.ars

    Given the amount of discussion on this forum about the power of network effects and Googles efforts to prevent the power of a monopoly to lock it out of the Net, readers here should immediately understand the threat of mergers that move us back in the direction of Ma Bell. I don’t think that even 3 providers is enough to provide adequate competition, and would love to see more, not fewer. I hope that the merger is blocked, and that T-Mobile can find a way forward without merging with any of the Big 3.

    What would smartphone prices look like if, say, Apple and HTC had 85% market share between them? I don’t think it would be pretty. The provider situation is even worse, because new entrants can manufacture phones more easily than new entrants can obtain RF licenses and build out infrastructure.

  49. >I don’t think that even 3 providers is enough to provide adequate competition, and would love to see more, not fewer.

    I agree. My libertarianism leaves me somewhat conflicted here; I don’t want the merger to go through, but I don’t trust the government with the power required to prevent it.

    I’ve thought through some post-merger scenarios, and I think that if it goes through we are far more likely to see Google move to become a wireless ISP with its own infrastructure, perhaps by buying up Sprint’s. I don’t think Googles’s uberstrategy can tolerate the effect of a tight carrier duopoly, and I see nobody else with both the business motive and the financial mass to take that on.

  50. @Jeremy Bowers:

    > I can talk out my own ass as well as any other not-an-EE.

    The thing is that we’re all talking out our asses to some extent. Building a chip is a huge undertaking, and (with very few exceptions) practitioners are a lot like the blind men surrounding the proverbial elephant. Even though I work in a semiconductor company, I work on the system design side of things. We have to understand enough about what is going on to make rational decisions, but no matter what, we all wind up relying a lot on what others tell us.

    For example, the issue about finer geometries. I know a lot about the issues at a high level: the mask sets cost more, the wafers cost more, the defect density changes, transistor leakage and voltages change, etc. But what I don’t directly know is the gory details of actually applying the finer geometries, because I don’t work in a fab…

  51. @esr: “If [the AT&T/T-Mobile merger] goes through we are far more likely to see Google move to become a wireless ISP with its own infrastructure, perhaps by buying up Sprint’s…I see nobody else with both the business motive and the financial mass to take that on.”

    Microsoft.

  52. @esr: “My libertarianism leaves me somewhat conflicted here; I don’t want the merger to go through, but I don’t trust the government with the power required to prevent it.”

    My personal opinion is that libertarian-style competition can only thrive in an environment where anti-trust is enforced, or in industries that lack positive returns to scale. Without anti-trust, the biggest players will execute regulatory capture to lock out other players. Theoretically, a libertarian state would have no regulatory body to capture, but that appears to me to be an unstable equilibrium. The history of the last century, as you’ve clearly pointed out elsewhere, is an evolution away from night-watchmen states toward those with strong central authorities and ever-expanding political, economic, and regulatory powers. This occured because of the strong constituency that believed they would benefit from it.

    What does libertarianism suggest as a counterbalance to ensure that night-watchmen states remain small and inoffensive in the face of citizen demand to expand their powers?

  53. > My libertarianism leaves me somewhat conflicted here

    It is probably unknowable whether a competition-destroying duopoly could develop in a truly free market absent government intervention, but if it could, then whatever game ruleset is currently in use to define the “free market” would require modifications.

    So to the extent that the ideal government embodies a small, regular set of rules, it may always be necessary to have the big rule that says “we are going to step in and change things if these objective criteria indicate a lack of competition.” Of course, it would be best to tweak the other rules so that one seldom needs to be used, and it is pretty obvious that a lot of the rules that allowed monopolies and duopolies to develop were actively promulgated by the government.

    But given the lack of political will to fix or apply some of the laws in a fashion that would keep cartels from forming in the first place, I am personally more disturbed by the recent tendency of the courts to reduce the application of the antitrust laws than I am by the fact of the existence of said laws.

  54. >Without anti-trust, the biggest players will execute regulatory capture to lock out other players

    Funny, I consider that the usual effect of anti-trust law – that is, as a club wielded by the politically favored against those that free markets have rewarded.

    >What does libertarianism suggest as a counterbalance to ensure that night-watchmen states remain small and inoffensive in the face of citizen demand to expand their powers?

    Welcome to one of the principal reasons I’m an anarchist rather than a minarchist. :-)

  55. @ Patrick
    > I know a lot about the issues at a high level: the mask sets cost more, the wafers cost more, the defect density
    > changes, transistor leakage and voltages change, etc. But what I don’t directly know is the gory details of actually
    > applying the finer geometries, because I don’t work in a fab…

    I’m the Unix admin for a photomask company, so I’ve got a little extra insight here on the mask side from what I’ve gleaned from the engineers. The smaller your critical dimensions, the better quality your glass needs to be, so you _start_ with a 4 or 6 inch piece of quartz that costs thousands. Then you include your multi-million dollar lithography tool that depreciates over a Rather Long Time. Throw in the costs of chemicals for wash and resist and cleaning and so forth, plus all the IP secret-sauce that goes into each mask shop’s wet and dry processes, and then allow for a higher-than-normal defect rate due to the feature size, which potentially involves starting all over with a new plate of quartz. Technologies like Optical Proximity Correction and Phase-shift can help a lot, but then you’re getting into extra cost again.

    The site I sit at produces, at its high-end, masks for 65nm chips, and a mask set for that node costs hundreds of thousands. Data sets are usually 10 to 30 GB per layer (or more), and it can take a goodly chunk of a day to write to a single plate.

    And this is just for the photomasks!

  56. @jsk:

    Does optical proximity correction require action on the part of the photomask shop (other than providing raw parameters to its users)? I thought that was something that would be pre-applied by, e.g. Cadence software on the GDS deck, and didn’t think it was a real cost adder.

  57. @jsk:

    > And this is just for the photomasks!

    Don’t know what you mean “just”. AFAIK the two main reasons why semiconductor companies really believe in thorough pre-layout verification are (1) the cost of the mask set and (2) the time through getting the masks, through the fab, and through packaging, before you get the first parts back.

    Number (1) is a direct cost, but number (2) can lead to huge opportunity costs — if you can’t test anything until you have something physical back from the fab, you can easily chew up an extra year in half a dozen cycles. Taken together, these two costs mean that (a) chip companies are very careful about their designs, and (b) chip companies have gotten very good (often with some hooks in the original design) at “patching” where we just need to regenerated one or two metal interconnect masks instead of the whole set to make (in some cases) some fairly significant changes.

  58. @ Patrick:

    I say “just” because it’s only one part of the whole process, and I can’t speak at all for the other steps. Design is design, but I don’t know anything about the wafer fab processes. Just masks.

    As for OPC:
    The company I work for provides OPC services as a value-add. You might be thinking MRC or DRC, as they use similar software stacks and do not-dissimilar things. The software (whether Synopsis or Mentor or whoever) is awfully expensive as I understand, and every layer takes some fine-tuning when running it through OPC. MRC/DRC can be automated. I only understand the OPC tech peripherally; I support the systems. We’ve got a dedicated engineer team that knows the nitty-gritty.

  59. @Cathy:

    What does libertarianism suggest as a counterbalance to ensure that night-watchmen states remain small and inoffensive in the face of citizen demand to expand their powers?

    “The tree of liberty must be refreshed from time to time with the blood of patriots and tyrants. It is it’s natural manure.” — Thomas Jefferson

  60. “My personal opinion is that libertarian-style competition can only thrive in an environment where anti-trust is enforced, or in industries that lack positive returns to scale. Without anti-trust, the biggest players will execute regulatory capture to lock out other players.

    That bolded part is the issue. Libertarian-style competition does not allow regulatory capture. You really can’t find very many examples of monopoly that do not involved government coercion of some form to maintain.

  61. > Without anti-trust, the biggest players will execute regulatory capture to lock out other players.

    Hmm. The US has both anti-trust and regulatory capture. So, if it’s also true that no-anti-trust leads to regulatory capture, anti-trust is irrelevant to regulatory capture.

    Or, do you want to argue that we should ignore the regulatory capture that we have with anti-trust? How about not? How about we wait until you manage to get rid of regulatory capture in an environment that has anti-trust before we act as if that’s true.

  62. @Cathy: “My personal opinion is that libertarian-style competition can only thrive in an environment where anti-trust is enforced, or in industries that lack positive returns to scale. Without anti-trust, the biggest players will execute regulatory capture to lock out other players. ”

    @SPQR: “That bolded part is the issue. Libertarian-style competition does not allow regulatory capture. You really can’t find very many examples of monopoly that do not involved government coercion of some form to maintain.

    I understand that that’s the theory. But how do you keep libertarian-style competition from degrading when the citizens/voters/gov’t etc. decides that it really needs regulation, and suddenly you have something worth capturing? How do you stabilize a libertarian system so it doesn’t drift toward an authoritarian state by making promises to those who believe they’ll benefit from exercise of state power?

    Morgan Greywolf hints that it takes a revolution to put the mushroom cloud back into the uranium sphere, but that’s tantamount to stating that a libertarian system is unstable. Of course, one could easily make the argument that any government system is at best meta-stable. Heck, you can read Plato’s Republic and see that even the classical philosophers understood clearly that a government can degenerate over time if its citizens allow their ideals to degenerate. From an idealistic state to a rough-and-ready democracy to tyranny…

    “Our Constitution was made only for a moral and religious people. It is wholly inadequate to the government of any other.” –John Adams

    Perhaps one day we’ll discover how to make a government that encourages freedom and yet does not require its citizens to choose morality of their own free will. However, I would not suggest holding one’s breath waiting.

    I think I prefer ESR’s response — the system is fundamentally unfixable, so go for anarchy (i.e., no system at all). :-)

  63. @Cathy: “Without anti-trust, the biggest players will execute regulatory capture to lock out other players.”

    @Andy Freeman: “Hmm. The US has both anti-trust and regulatory capture. So, if it’s also true that no-anti-trust leads to regulatory capture, anti-trust is irrelevant to regulatory capture. Or, do you want to argue that we should ignore the regulatory capture that we have with anti-trust? How about not? How about we wait until you manage to get rid of regulatory capture in an environment that has anti-trust before we act as if that’s true.”

    I’d like to think that it’s possible to design a system that would block the concentration of power sufficient to oppress the liberties of others, hence no private entity powerful enough to do regulatory capture. But the existence of a power center strong enough to break up too-powerful entities implies some kind of regulatory authority. And of course that power center can become oppressive in its own right. Round and round we go…

    I’ve love to say I have an answer, but I don’t.

    “Anyone who thinks the United States has solved the problem of government is using too short a time scale.” –RAH

  64. @Cathy
    “I’d like to think that it’s possible to design a system that would block the concentration of power sufficient to oppress the liberties of others, hence no private entity powerful enough to do regulatory capture.”

    You are on your way to reinvent democracy.

    And if people really do not want libertarianism, it seems futile to me to try to force it on them.

  65. @ Winter
    > to force it on them.

    I do not think this word means what you think it means.

  66. @Winter: “You are on your way to reinvent democracy. And if people really do not want libertarianism, it seems futile to me to try to force it on them.”

    @jsk: “I do not think this word means what you think it means.”

    Indeed. This is actually a very fundamental point that goes beyond libertarianism, indeed to the essence of democracy and why it so often fails in cultures that do not have a democratic history. It also goes to why American democracy has often failed to give its benefits to everyone.

    There are two core principles at the root of a properly-functioning democratic state: (1) majority sets the rules, and (2) strong legal protections for minorities. (“Minorities” here refers to any groups that does not command a majority of voters, and does not have the same racially-based meaning it has come to have in modern America.)

    (1) is popular worldwide, and the strong support for democracy in authoritarian states comes from the desire of most people to have their wishes heard and the tyranny overthrown. Unfortunately, once a coup or revolution throws off the authoritarian state, it is often replaced by an oppressive democratically-elected state.

    That occurs because most people do not really understand part (2), the protection of the rights of citizens who are not members of the majority think-alike group. Tutsis up, Hutus down; Hutus win a war, and promptly turn around and start killing Tutsis. New democracies really struggle with the concept of a loyal opposition, and the need to have power alternate between two or more parties. (The real acid test of the American Revolution was not 1776 or 1783, but 1800, the year that the Democrat-Republicans peacefully took power from the Federalists after a campaign that was very bitter on both sides.)

    Libertarianism is really the extreme case of (2), taken to the mathematical limit; if you do not want a service, you cannot be forced to accept that service and cannot be required to pay for others to get it either. But because all democracies struggle to balance (1) with (2), the right to opt out gets eroded very easily. In the U.S. today, (2) is very weak for members of groups that are not “recognized” minorities.

  67. >I do not think this word (“force”) means what you think it means.

    Remember that Winter has described himself as a communist. It follows that he is a moral idiot who is either incapable of reasoning about the ethical use of force, or willing to lie about his actual position in the service of totalitarianism, or both. Treat any of his comments on political or ethical matters with the same sort of skepticism you would apply to discussion of racial tolerance by a Nazi.

  68. And if people really do not want libertarianism, it seems futile to me to try to force it on them.

    Robbers and rapists don’t want libertarianism (punishing aggressors), but I have no qualms about “forcing it on them”.

  69. “>What does libertarianism suggest as a counterbalance to ensure that night-watchmen states remain small and inoffensive in the face of citizen demand to expand their powers? Welcome to one of the principal reasons I’m an anarchist rather than a minarchist. :-)”

    You want libertarianism, “most” (as in a qualified majority) seem to prefer an expanded state power. What to do? Prevent the reconstitution of a state? How?

    Many here fulminate against statists. But they seem to form a heavy majority. They do not want to live in a libertarian state I presume. So I was saying that forcing them to do so seems futile to me. As in, a lost case.

    Not suggesring you wanted to use violence against them

  70. Libertarian-style competition does not allow regulatory capture.

    capital == power

    Libertarian arrangements are precisely those which enable the greatest regulatory capture by those with the most accumulated capital. Regulation is force; absent the state, force is just another commodity to be bought and sold.

    Why TF do you think the Kochs bankrolled the LP and various libertarian/right-wing think tanks?

  71. @Jeff Read: Why TF do you think the Kochs bankrolled the LP and various libertarian/right-wing think tanks?

    I don’t know, Mr. Mind Reader, why do they do this?

  72. No, force is not a commodity that can be bought and sold. Mercenaries aren’t idiots; they quickly figure out that it’s easier to get the money by aiming their guns at the rich guy than by fighting the targets named by the rich guy.

  73. capital == power

    It might seem that way, for someone with a severe money fixation. But the reality is that money alone doesn’t buy power, although it does help.

  74. @Jeff Read, Libertarian policies don’t permit the regulatory agency to exist that is required for regulatory capture. QED.

  75. Ah, I know the implied argument that I missed. The argument is that if we move from a statist level of regulating a market at “11” down to “8”, then the fact that there is still regulatory capture is the fault of the “libertarian” policy that reduced the statism by a fraction.

    Sorry, but I can’t take that argument seriously.

  76. Android news du jour:

    digitimes (which seems to have direct knowledge of stuff that goes on in the Asia electronics community and historically has pretty good accuracy) claims that HTC is giving Apple a run for its money in the US this quarter. Makes good sense, particularly since most people wanting an iPhone are now waiting for iPhone.next, whatever that happens to be.
    Analyst surveys confirm that Android is catching up to Apple at the retail level. I suspect (from previous surveys vs. previous sales) that Android has already historically been far, far ahead in online sales.

    Motorola is aiming squarely at RIM. You know you’re in trouble when the Android also-ran is gunning for you with live ammo. This also has the (intentional) side effect of entrenching Mot at Sprint before Apple gets there.

    IDC is still claiming that NoWin will eat Apple’s and RIM’s lunches.

    Sprint has a trade-in program that apparently let at least one EVO 4G user upgrade to a Nexus S for $50.

    I don’t know how real PhonesOverstock.com is, but they’re supposedly offering the as-yet-unreleased HTC Sensation 4G for $160 with no contract.

    The Samsung Infuse 4G is available for new AT&T customers for $119 at Amazon.

    And, of course, beatweek magazine is still at it. They’re shocked! that Android is doing so well, and sincerely hope that Apple will recover before the year is out.

  77. (2) is very weak for members of groups that are not “recognized” minorities.

    True, and it’s actually worse than that: according to federal civil rights law, Jews do not count as an “historically oppressed minority.”

  78. ‘…according to federal civil rights law, Jews do not count as an “historically oppressed minority.”’

    Hey…I’m Jewish, and while I and other Jews I know have experienced isolated incidents of antisemitism in America, the fact is that American Jews (as a group) are better educated and make more money than the national average. Special legal protections of that type aren’t necessary.

  79. A cheap SOC that use a different instruction set and is funded by the Chinese????

    Hahahaha….. to me spells doom for Google than Apple!!!!

    No wonder Google is now lately been very hesitant to release it codebase for android Honecomb or any future android release to the opensource community.

    How long has HoneyComb been release now??? Is it nearly 6 month already? Was it early February?

    Why is it afraid somebody else may hijack your android OS now the Android has manage to be so popular?

    Goggle…Oh google… that is what happen when you piss of the Chinese!

  80. American Jews (as a group) are better educated and make more money than the national average. Special legal protections of that type aren’t necessary.

    So that makes it OK to discriminate against you, but not members of groups that earn below average? It means that every time someone has to decide what sub-contractor to hire to get a job done, he should ignore the Jewish-owned company, and give his business to the black paraplegic lesbian from Puerto Rico, even if the latter charges twice as much money and does crappy work?

  81. Speaking of fact checking, can we get a citation on the loosely phrased “according to federal civil rights law”?

  82. I don’t actually think so. As long as they can port Dalvik, who cares what the instruction set is?

    Anyone who cares about performance cares about the instruction set.

    Cellphones turn out to have a lot of applications that need some pretty serious computing performance: high-speed games, DSP applications (a lot of cellphone apps fit in here, everything from rangefinders to VoIP clients to “T-Pain” vocal processing toys), visualizers and AR applications, things that may require (soft) real-time control, etc.

    Google have helpfully provided a native API for this sort of thing. Most apps that use it have native-code modules compiled to the ARM platform because that’s what the vast bulk (as in all) of smartphones use. Now you want to add a whole ‘nother instruction set into the mix? Can you say FRAGMENTATION?

    Really, though, the Android strategy was fukt to begin with, even if you don’t add in the IP issues from Oracle, Apple, and Microsoft. Smartphones are much more resource-constrained than modern PCs (though they’re at or better than a PC from, say, a decade ago). What’s more, they can’t be upgraded. You put an atificial sophistication constraint on your application by adopting a VM/GC based runtime. What should have been done is to use native code throughout, with objects on the heap managed through reference counting.

    Sounds a lot like iOS doesn’t it?

    Turns out there’s a lot that Apple got right the first time that Google is still struggling with.

  83. Jeff Read, I don’t see how you can credibly claim that IP issues are unique to Android. Apple’s iOS “strategy” is no more immune to lawfare than Android. E.g., Lodsys.

  84. “It means that every time someone has to decide what sub-contractor to hire to get a job done, he should ignore the Jewish-owned company, and give his business to the black paraplegic lesbian from Puerto Rico, even if the latter charges twice as much money and does crappy work?”

    Actually, she’s a paraplegic because she’s a disabled Vietnam veteran, whose mother took DES before she was born….Oh, and she happens to be Jewish. (You should hear her sing La Borinquena in Yiddish. Connie Francis had nothing on her.)

    Coming back to reality, I agree that there is a lot of unfairness in minority contracting set-asides. A lot of those ‘minority businesses’ are just shells around majority-type owners, some of whom are Jewish.

  85. JonB…

    Maybe you just get a C??? for your facts!!! Mr. B….

    Tell us when honeycomb will be open source?
    It is now reaching 5 month….

    What r reason why HoneyComb is tide to the Tergra 2 Processor/SOC???

    Is it to avoid FRAGMENTATION????
    Maybe it is also so a fork can never happen…
    and everyone is tide to google solution/services??

    …………….

    There just damn tooo many GPU out there and they r not compatible to each other!!!

    Wonder why there is no GPU Acceleration on ANDROID???
    Why google prefer the stupid Dalvic/Java solution rather than Native solution?

    Shhh……
    it’s fragmentation….
    must be really cheap to solves all this variation….

    ………………………………………………

    What is Open will seek to Close!!!
    What is Close will seek to Open!!!

    The chinese has manage to fork android with thier Tapa OS…

    Won’t everyone else want their own version of android?

    Won’t the sony guys want a Playstation Phone that plays their Game only on sony’s hardware?? AND LATER release some different version for the rest of hardware?

    Won’t facebook want an Android phone that uses mostly uses Facebook’s services?

    what about Amazon??? what about yahoo??? what about the phone makers and chp maker???What about even Microsoft???

    Android is there for the taking…..

    ………………………………………………………………………….

    so with all this happening, is android really in trouble???

    Android was never in trouble from the beginning… it is opensource just like linux and Meego. Android would carry on.

    Fragmentation is not a problem for Android….

    Fragmentation is a problem for Google!!!!!

    They can’t make money out of it while it is insanely popular/available everywhere.

    …………………………………………………………………….

    Google after being suckerpunch JOBS
    with is Free and AD FREE iCloud services plus Twitter partnership.

    Making google’s own Cloud and other services a lost revenue.

    And improving both the notification and widget part of IOS over Android with Notification center for IOs.

    while Google’s heavily push the stupid NFC payment scheme for Android while failing to improve the Gingerbread ANdroid over Froyo.

    and still failing to get the deal done with the music services/store while claiming they these guys need android more than android needs them.

    ………………………………………………………………………

    We will continue this when we we see the opensource Honeycomb…

  86. @kk man:

    1) You keep capitalizing AD FREE as if the average consumer actually cares about it. If ad free were really that important, pay-per-channel television would have eaten broadcast and basic cable alive. Think about it.

    2) Fragmentation doesn’t matter for Google at all. The purpose of Android is drive ad revenue from Google’s bread and butter search business. Fragmentation isn’t a problem there because people will continue to use Google’s search engine even if they aren’t using an official Android build.

    3) Gingerbread was actually a big improvement over Froyo, at least for me. Battery life and memory management have improved greatly. My EVO 4G will actually go about 3 days without a charge, so long as I keep my usage relatively light. That’s almost as good as my old feature phone.

    4) Music store — music stores don’t matter for the large silent majority still getting their music from streaming services, CD ripping, and copyright infringement.

    Try doing a little research before you speak.

  87. Two points:

    >You keep capitalizing AD FREE as if the average consumer actually cares about it. If ad free were
    >really that important, pay-per-channel television would have eaten broadcast and basic
    >cable alive. Think about it.

    Pay per channel is still not free. The important combination is free as in beer* and ad free as well. And in any situation where consumers can choose equivalent content and service without ads for free, they always choose “without ads”. For evidence of this, look no further than the popularity of ad blockers. You can also look at the days way back when you could get dial up for free if you would look at ads, people did everything they could to avoid the ads, even though that was part of the service agreement.

    >4) Music store — music stores don’t matter for the large silent majority still getting their
    >music from streaming services, CD ripping, and copyright infringement.

    The store may not matter to them for buying new music, but what about this new “match” thing? If it works as promised, that’s a hell of a convenience for that large silent majority. No, the guys downloading from the napster spawn likely aren’t going to pony up for it, but the guys with large CD collections?

  88. SPQR Says:

    > Jeff Read, I don’t see how you can credibly claim that IP issues are unique to Android.
    > Apple’s iOS “strategy” is no more immune to lawfare than Android. E.g., Lodsys.

    When I read this, I thought ‘Nobody is immune from a patent troll, because patent trolls are immune to the nuclear option’, but I assumed Apple’s patent portfolio better protected them from competitors until I read this:
    http://apple.slashdot.org/story/11/06/14/1230228/Apple-Agrees-To-Pay-Licensing-Fees-To-Nokia

  89. Apple has apparently settled in their multi cross patent lawsuit with Nokia. here

    One wonders how much that is because they think their legal case is weak, and how much it is a move against Android which, with its open source model, will have a real difficult time making a similar settlement.

    Call me a cynic? Maybe. Call me a hater of all things patent? Definitely.

  90. In other news, Apple is officially selling the iPhone unlocked, albiet for even more than the original iPhone cost. http://store.apple.com/us/product/MC603LL/A

    Bets on how long before the price drops, and subsequently how long it takes for people to get up in arms and demand apple give them free money for dropping the price.

  91. @Jessica Boxer:

    Google would have no trouble at all settling with Oracle as all of the code in question is available under the terms of the Apache license, which lacks the GPL’s restrictive anti-patent clause. Google is not fighting Oracle because of any perceived lack of choice due to its open source model.

    The Microsoft settlement isn’t problematic because Microsoft has wisely chosen to not restrict anyone’s right to redistribute the Linux kernel. (Because the second they did this, an all-out patent assault on Microsoft would likely break out, with IBM and its nazgul in the lead.)

  92. @tmoney:

    And in any situation where consumers can choose equivalent content and service without ads for free, they always choose “without ads”.

    I didn’t have to say a thing.

    The store may not matter to them for buying new music, but what about this new “match” thing?

    I have my own “in-the-cloud” storage – 1.5 TB that I can access from anywhere in the world via my mobile phone, a Web browser, or any scp client. What does this service offer me? Almost nothing.

  93. >I didn’t have to say a thing.

    Once again, pedantry is not a substitute for an actual argument.

    >I have my own “in-the-cloud” storage – 1.5 TB that I can access from anywhere in the world via
    >my mobile phone, a Web browser, or any scp client. What does this service offer me? Almost nothing.

    Are you talking about iCloud as a whole, or specifically the match service?

  94. I should also note that even if the service offers you “almost nothing” you are hardly a representative consumer, and certainly not a representative consumer of Apple products. It never ceases to amaze me how difficult this concept is for techie people to grasp. Apple succeeds because they really do try (and often succeed) at making the “it just works” concept true, because most people really don’t have the time, inclination or knowledge to set up and work the things that we find so easy.

  95. @tmoney
    > It never ceases to amaze me how difficult this concept is for techie people to grasp.

    Ideology aside, I think it’s more an issue of two groups talking at odds. Seems to me ‘techie’ people are often specifically responding to the (not always) mild marketing dishonesty that says Apple is always trailblazing entirely new and unimagined technology. [Sensible] Apple fans retort with the real points of interest (usability, interface, polish) without actually addressing what the ‘techie’ people are complaining about.

    I have no use for a lot of Apple’s stuff, but they do what they do awfully well. It does bug the hell out of me that they, and their fanfolk, are such prats about it. Yes, your UI is nice, yes, you did a good job. No, the actual technology is nothing special, now quit acting like it is.

  96. > I have my own “in-the-cloud” storage – 1.5 TB that I can access from anywhere in the world via my mobile phone, a Web browser, or any scp client. What does this service offer me? Almost nothing.

    BFD. Many people who read/post here have similar “personal clouds”. That’s not the point. Eric’s Aunt Tillie doesn’t want to have a machine (or VM) in colocation, doesn’t want to pay the bills for same, etc. Nor does she want to think of the issues with having N copies of M documents all in flow between different devices, etc. She doesn’t want to run ‘git’ to keep changed managed, (nor do you, apparently).

    She doesn’t want to think about it, she wants it all to “just work”.

    that is advantage #1 of iCloud.

    Advantage #2: it’s free, with zero ads.

    Advantage #3 is that Apple’s “Back to my Mac” is part of iCloud, and with it’s ability to securely(*) penetrate the NAT devices at the perimeter of nearly every home, meaning that we’re going to be very, very close to everyone with a Mac and an iPhone having a personal cloud. I see your 1.5TB and raise you.

    * Kerberos-authenticated, IPV6 tunnels secured with IPSEC, directly back to the Mac, with facilities for supporting waking the machine from sleep on access from the remote. (Bonjour Sleep Proxy) Oh, and all the technology is either based on open source (Raccoon, Kerberos) or have been open sourced by Apple (Bonjour, Bonjour Sleep Proxy)

  97. @not(Andy Rubin):

    You sure sound like an ad for Apple. Others have already told you that it’s extremely easy not to have ads on google properties. I don’t recall the last time I’ve seen one, and I don’t even know that I do anything special.

    In other news, AMD has seen the mobile light and started focusing on battery life. Their first effort is now available:

    http://www.geekwire.com/2011/amds

  98. @tmoney
    “Apple succeeds because they really do try (and often succeed) at making the “it just works” concept true, because most people really don’t have the time, inclination or knowledge to set up and work the things that we find so easy.”

    That is because Apple take existing technology, and now services, repackage them zero-conf, and market them. Their offerings are not the first, not the cheapest, and not the best. But they sure know how to package and sell them better than anyone else.

    And that is the point. Most “techie” people consider other things important than the target audience of Apple. I can understand very well why some people like Apple products. I have no idea why people who like Apple products cannot understand that there are many people in the world who actually have different preferences and needs. On the other hand, I just have to listen to the next religious type to observe the same.

  99. @tmoney: I should also note that even if the service offers you “almost nothing” you are hardly a representative consumer, and certainly not a representative consumer of Apple products.

    Pot, kettle? You are saying:

    @tmoney: For evidence of this, look no further than the popularity of ad blockers.

    Well, let’s take a look, shell we? It’s not that hard.

    Ad blockers: https://addons.mozilla.org/statistics/addon/1865 – about 14million users
    Internet users: http://en.wikipedia.org/wiki/Global_Internet_usage – about 2 billion users

    If we assume Firefox users are at 20% (conservative estimate) then that means about 4% of users use Ad blockers (well, may be 5% if you include less-popular Ad blockers).

    So… AD FREE is really big advantage. If all geeks who care about them will boycott Android (every 20th mobile user, remember), then… Google will hardly notice the dip. Q.E.D.

    Note that there are 124838419 downloads of AdBlock yet 14063537 weekly users. This means that people who care about AD FREE… (you are among them, right?) they really do care. And are ready to go to great lengths to block them. But there are so, sooo, few of them. It never ceases to amaze me how difficult this concept is for techie people to grasp.

    Forget about AD FREE. This horse is already dead. Was dead eons ago, in fact.

  100. @not(Andy Rubin)

    iCloud may work for your definition of “just work”, but your definition isn’t the only one.

    iCloud stops working the minute you use even one non-Apple mobile device (Android, Kindle, webOS, WP7, Symbian, BB, QNX, …).

    iCloud stops working the minute you lose your iPhone (or iPad) and doesn’t start working again until you replace it.

    iCloud’s sync-focused architecture doesn’t “just work” if you’re looking at a space-constrained device (if you were trying to design a really cheap iPhone, for instance…).

    iCloud doesn’t “just work” if you have more than one “Apple identity” (e.g. a MobileMe email for business use and an iTunes account for personal use).

    And so on…

  101. Hey, khim — did you know that Firefox, Safari, IE, Chrome, and Opera all come with popup blockers built in?

  102. @not(Andy Rubin):

    Advantage #3 is that Apple’s “Back to my Mac” is part of iCloud, and with it’s ability to securely(*) penetrate the NAT devices at the perimeter of nearly every home, meaning that we’re going to be very, very close to everyone with a Mac and an iPhone having a personal cloud. I see your 1.5TB and raise you.

    * Kerberos-authenticated, IPV6 tunnels secured with IPSEC, directly back to the Mac, with facilities for supporting waking the machine from sleep on access from the remote. (Bonjour Sleep Proxy) Oh, and all the technology is either based on open source (Raccoon, Kerberos) or have been open sourced by Apple (Bonjour, Bonjour Sleep Proxy)

    I have that, too. It’s called OpenSSH and/or OpenVPN, combined with the Wake-On-LAN functionality built into my DD-WRT-based router and el-cheapo Realtek chipset onboard NICs.

    Oh, yeah, and I paid $0 for my operating system (Ubuntu 10.04.2 LTS, currently)

  103. >Pot, kettle?

    Not at all, I am fully aware that I am not representative of Apple’s general consumer market. Or Dell, or HP, or any of the major computer manufacturers. I would also wager that 14 million ad block users is significantly larger than the number of people who have their own personal clouds right now.

    >iCloud stops working the minute you use even one non-Apple mobile device
    >(Android, Kindle, webOS, WP7, Symbian, BB, QNX, …).

    Yes, one of the conceits of Apple’s “Just Works” philosophy is that you have to use Apple’s stuff to get it to “Just Work”. You will also find that OS X’s features that “Just Work” don’t work if you don’t use OS X.

    That said, the iCloud will have an API for developers so presumably at they very least some of these devices can be made to work in the future.

    >iCloud stops working the minute you lose your iPhone (or iPad) and doesn’t start working
    >again until you replace it.

    I’m fairly certain that it keeps on working. Yes, you can’t get it from anywhere, but then again, without your mobile device, you can’t get anything from just anywhere anymore. This is like me saying that Gmail stops working if I lose my Android phone.

    >iCloud’s sync-focused architecture doesn’t “just work” if you’re looking at a space-constrained
    >device (if you were trying to design a really cheap iPhone, for instance…).

    How does it not work? Swap something off the device, or even stream it directly to the device. Further, how is this any different with a cloud than with regular old wired syncing?

    >iCloud doesn’t “just work” if you have more than one “Apple identity” (e.g. a MobileMe
    >email for business use and an iTunes account for personal use).

    I can’t speak to the specific iCloud implementation since it isn’t out yet, but I’ve used multiple MobileMe accounts with the same device just fine. On what basis do you assume this won’t work?

    It seems to me that you are seriously reaching here for anything you can use, no matter how tenuous.

    >I have that, too. It’s called OpenSSH and/or OpenVPN, combined with the Wake-On-LAN
    >functionality built into my DD-WRT-based router and el-cheapo Realtek chipset onboard NICs.

    Ladies and Gentlemen, the prosecution rests.

  104. How is it pedantry to point out that the type of person who might setup an OpenSSH with OpenVPN DD-WRT router with a Wake-On_LAN setup to his Realtek chipset computers might be outside general target market for Apple?

    In my tenure at Apple most of the customers I dealt with couldn’t tell you what a LAN was, let alone what any of those other terms might mean. And of the ones that could, I would venture that less then half of those could tell you what chipset their NIC used.

  105. How is it pedantry to point out that the type of person who might setup an OpenSSH with OpenVPN DD-WRT router with a Wake-On_LAN setup to his Realtek chipset computers might be outside general target market for Apple?

    KMy point is that Apple can’t claim that that iMatch is new tech. It isn’t. Ever heard of rsync? All Apple has done is packaged something integrated, shiny and pretty. That’s not new, and, in fact, I have created such pretty setups for my own customers. What difference does it make whether someone pays me or pays Apple? None, except to Apple, of course.

    As for pedantry, specifically, I was referring to your comment: “I’m fairly certain that it keeps on working…etc.”

    That’s pedantry.

    In my tenure at Apple most of the customers I dealt with couldn’t tell you what a LAN was, let alone what any of those other terms might mean. And of the ones that could, I would venture that less then half of those could tell you what chipset their NIC used.

    Sure. That’s how make my money.

  106. @tmoney

    First, I’d like to be very clear on what I am arguing here. I agree that Apple’s definition of “Just Working” is understandable and internally self-consistent. I merely claim that Apple does not have the One True Vision of things “Just Working”.

    Specifically, I am arguing that there are other equally reasonable definitions of “Just Working” that iCloud (in combination with other Apple products and services) does not meet and that alternative products and services (usually Google’s ) do meet.

    Let’s take your comments point-by-point

    > iCloud will have an API for developers so presumably at they very least some of these devices can be made to work in the future

    There’s a difference between developer APIs for iOS / OS X apps and developer APIs for non-Apple devices. I would be very surprised if there were non-trivial iCloud APIs for anything other than iOS, OS X and Windows, but that’s a side point. My main point is that while it is reasonable for Apple to insist that things are only going to “Just Work” if you only buy their devices, it is equally reasonable for me to insist that your product or service “Just Works” on any feasible device I pick (see, for example, Amazon and Kindle books).

    > This is like me saying that Gmail stops working if I lose my Android phone.

    As long as you can get to an Internet cafe / friend’s computer / …, Gmail, like most Google services, keeps working very, very well even if you lose your Android phone. By contrast, iCloud will not work well (if at all – see, for example, music downloads) until you acquire a new, *personal* device. Requiring a personal device is reasonable, but it is equally reasonable to insist that a personal device shouldn’t be required.

    > How does it not work? Swap something off the device, or even stream it directly to the device.

    It seems you have missed this detail: Despite all the pre-WWDC rumors, iCloud does *not* stream music. You can download / sync music to any of your N devices, but iCloud will not stream it for you (at least for now). My whole point was that on a space-constrained device (e.g. 1 GB or smaller) swapping isn’t a usable “Just Works” option. If you disagree, I’d invite you to take the T-Mobile G1 challenge (which I recently did while waiting for my G2 to be replaced). Requiring a device with 8 GB+ of local storage is reasonable, but it is equally reasonable to say that you want some basic functionality as cheaply as you can get it (implying, among other things, as little space as you can get away with).

    > I can’t speak to the specific iCloud implementation since it isn’t out yet, but I’ve used multiple MobileMe accounts with the same device just fine. > On what basis do you assume this won’t work?

    See: http://www.macsparky.com/blog/2011/6/6/icloud-and-multiple-itunes-accounts.html and http://forkbombr.net/icloud-and-apple-ids/
    (via http://theorangeview.net/2011/06/iclouds-one-login-jumbles-identity/ ). I’m not suggesting that these problems are insoluble (my understanding is that Apple support can smooth over at least some of this), but this is clearly not “Just Working”.

    By contrast, when Android didn’t support multiple Google accounts to begin with, people complained and Google fixed it. Saying that your personal device is only going to “Just Work” if you only have one associated identity is reasonable, but it is equally reasonable to say that I have multiple identities that matter (work, professional, personal, [extended] family, etc.) and my mobile device should integrate them.

  107. How is it pedantry to point out that the type of person who might setup an OpenSSH with OpenVPN DD-WRT router with a Wake-On_LAN setup to his Realtek chipset computers might be outside general target market for Apple?

    Here’s pedantry for you:

    The verb phrase is “set up”. The noun/adjective is “setup”.

    “John set up a DD-WRT router by connecting his web browser to the setup screen, where he uploaded a file off his laptop containing a saved setup from the previous router.”

    The same general rule applies to many similar words like “shutdown”, although in some cases, the noun/adjective form is hyphenated.

    This has been your Grammar Nazi Moment.

  108. >I agree that Apple’s definition of “Just Working” is understandable and internally self-consistent. I merely claim that
    >Apple does not have the One True Vision of things “Just Working”.
    >Specifically, I am arguing that there are other equally reasonable definitions of “Just Working” that iCloud (in combination with other
    >Apple products and services) does not meet and that alternative products and services (usually Google’s ) do meet.

    In that case, you will find no disagreement from me. Any broad stroke has its caveats, including Apple’s “It Just Works”. Since we’re in agreement on that point, I think both points 1 and 3 are mostly covered by the caveat that “It Just Works” applies specifically to Apple’s eco system and the limited third party devices covered therein. As specifically to the streaming portion, again I am assuming as iCloud is not yet released, but I was assuming the iTunes sharing would work with iCloud as it currently does with mobile me:

    http://forums.macrumors.com/showthread.php?t=783758

    Though I do admit that those links and references no longer appear on Apple’s website, and it apparently is not a well known feature as there aren’t many references to it in the wild web. However, as of 4.3 it is still present in iOS:

    http://www.pcworld.com/article/221589/how_to_enable_itunes_home_sharing_in_ios_43.html

    As to point 2, I think this falls under the difference between Apple’s approach to the cloud and Google’s. To Google, the cloud is the computer, to Apple the cloud is the storage. The converse of your argument is that without an internet connection, I can’t use most of Google’s apps or access my data unless I explicitly cached a copy locally. Neither approach is wrong as you say, but they each have their individual tradeoffs.

    As to point 4, I will admit I had not seen that the current iCloud setups only allow for a single account. I had again assumed that since multiple accounts works now, that it would continue to work in the future. I think if it isn’t available at launch, it will be shortly, because as you say, many people use it, and for them, iCloud would not “just work”

  109. @tmoney

    My (expanded) thoughts on iCloud streaming:

    1. Central iCloud streaming is clearly something Apple isn’t willing to promise today, since they didn’t announce it.

    2. For reasons we all understand, Apple is philosophically inclined against streaming (particularly streaming to non-personal devices).

    3. While I did miss that Apple already supported in-home streaming (good catch, BTW), in-home streaming doesn’t address the as-cheap-as-possible smartphone use-case.

    4. It isn’t realistic to get “It Just Works” mobile streaming by exposing the existing in-home streaming, even though many pre-WWDC rumors pointed in that direction. I can give you my full argument if you want, but here is the short version:
    (a) Apple would have to transparently juggle a long chain of moving parts (some of which they don’t control and some they would even find it difficult to influence).
    (b) Even if Apple succeeded with their juggling, most residential ISPs aren’t up to the job of being large-scale streaming sources, no matter how badly they might want to be.

    5. I think points 1-4 actually don’t matter because the real “price” of iTunes Match wasn’t the money, but rather Apple promising some variant “no streaming” (no free streaming? no central streaming? …). Why do I think that? Because both Google and Amazon would have been fools to turn down “Google / Amazon Match” unless there was some fundamentally unacceptable string attached. A condition on streaming seems to be the most likely candidate because streaming the biggest feature Google and Amazon have in common that Apple doesn’t have.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>