In a day or two I plan to do a blog post on the way engineering decisions can be critically important at cusp points in the history of society. In order for part of the argument in that post to make sense, my readers need to hear a story I’ve been hinting at in comments for some time: the first time that I personally made a difference in the world at what I believe was history-changing scale, and how that happened.
I think it was at the 1983 Usenix/UniForum conference (there is an outside possibility that I’m off by a year and it was ’84, which I will ignore in the remainder of this report). I was just a random young programmer then, sent to the conference as a reward by the company for which I was the house Unix guru at the time (my last regular job). More or less by chance, I walked into the meeting where the leaders of IETF were meeting to finalize the design of Internet DNS.
When I walked in, the crowd in that room was all set to approve a policy architecture that would have abolished the functional domains (.com, .net, .org, .mil, .gov) in favor of a purely geographic system. There’d be a .us domain, state-level ones under that, city and county and municipal ones under that, and hostnames some levels down. All very tidy and predictable, but I saw a problem.
I raised a hand tentatively. “Um,” I said, “what happens when people move?”
There was a long, stunned pause. Then a very polite but intense argument broke out. Most of the room on one side, me and one other guy on the other.
OK, I can see you boggling out there, you in your world of laptops and smartphones and WiFi. You take for granted that computers are mobile. You may have one in your pocket right now. Dude, it was 1983. 1983. The personal computers of the day barely existed; they were primitive toys that serious programmers mostly looked down on, and not without reason. Connecting them to the nascent Internet would have been ludicrous, impossible; they lacked the processing power to handle it even if the hardware had existed, which it didn’t yet. Mainframes and minicomputers ruled the earth, stolidly immobile in glass-fronted rooms with raised floors.
So no, it wasn’t crazy that the entire top echelon of IETF could be blindsided with that question by a twentysomething smartaleck kid who happened to have bought one of the first three IBM PCs to reach the East Coast. The gist of my argument was that (a) people were gonna move, and (b) because we didn’t really know what the future would be like, we should be prescribing as much mechanism and as little policy as we could. That is, we shouldn’t try to kill off the functional domains, we should allow both functional and geographical ones to coexist and let the market sort out what it wanted. To their eternal credit, they didn’t kick me out of the room for being an asshole when I actually declaimed the phrase “Let a thousand flowers bloom!”.
(I wish I could remember who the one guy who immediately jumped in on my side was. I think it was Henry Spencer, but I’m not sure.)
The majority counter, at first, was basically “But that would be chaos!” They were right, of course. But I was right too. The logic of my position was unassailable, really, and people started coming around fairly quickly. It was all done in less than 90 minutes. And that’s why I like to joke that the domain-name gold rush and the ensuing bumptious anarchy in the Internet’s host-naming system is all my fault.
It’s not true, really. It isn’t enough that my argument was correct on the merits; for the outcome we got, the IETF had to be willing to let a n00b who’d never been part of their process upset their conceptual applecart at a meeting that I think was supposed to be mainly a formality ratifying decisions that had already been made in working papers. I give them much more credit for that than I’ll ever claim for being the n00b in question, and I’ve emphasized that every time I’ve told this story.
Now look at what we avoided. With the chaos came a drastically decreased vulnerability to single-point failures – and I mainly mean the institutional kind, not the technical kind. If DNS addresses had been immobile and tied to individual legal jurisdictions, there would have been a far stronger case for keeping address issuance under government auspices (it’s not like phone numbers, which started out being privately issued on privately-owned infrastructure). The bad consequences would only have started with expense and bureaucratization; more than likely control of the Internet address space would have become, in many jurisdictions, a political instrument used to reward approved parties and control information flow.
Not only would this have been a bad thing in itself, it would have set a negative precedent for centralizing other aspects of Internet governance that might well have hardened into stone by the time the technology was ready for the general public a decade later. Those of you who think ICANN and fly-by-night registrars are a pain in the ass aren’t wrong, but you should be on your knees in thanks that we got a system with that much play and polycentrism in it. We very nearly didn’t.
If I were a dimwitted egotist I’d claim to have saved the world, maybe. But what really mattered is that I threw my disruption into a roomful of hackers with an innate distrust of hierarchy and a strong allergy to system designs with single-point vulnerability. It was that shared culture that made the difference, more than me – it produced my objection, and it produced people ready to hear it.
Still. It was my hand that went up. Anybody who grokked that computers could be mobile could have changed history just then, but it happened to be me, and it was my first time.
UPDATE: It’s been pointed out that while the RFC-issuing group that became the IETF existed in 1983-1984, it did not formally constitute itself as the “Internet Engineering Task Force” until 1986. So my use of the term here is anachronistic.
This was definitely an interesting history. You were in the right place, in the right moment… But everyone could change history, the problem is that a lot of people just don’t care.
Thanks for your hand raising. I’d hate to have a .es (or .cat) domain: I don’t want to be tied.
Aw, any military brat annoyed with changing phone numbers every PCS would have said the same thing.
They did listen. That is wondrous.
Certainly they listened, and that is nothing short or remarkable. Perhaps it is fair to say that what caused this moment of change was not so much the smart kid, but rather the culture of the Internet that meant that a smart kid had a voice. Which is to say, in the hacker culture smart trumps seniority, and that is rare indeed.
However, I have a question for the anarchist in Eric. Why have this centrally decided at all? The centralization of the DNS has lead to all sorts of problems. I see no reason why the IETF could not simply have specified a protocol for name resolution and allows customers, with the help of their ISPs to choose their own domain name resolution techniques.
What we have ended up with instead is a situation where, because of the very slow process to approve top level TLDs, a culture has set in that .com is the only one worth having, and now a situation where it is close to impossible to obtain a useful, and usable domain name. The one that really made me laugh in this regard is “goarmy.com”, the US Army’s recruiting site. The US Military has a whole TLD of their own, but they still feel the need to use .com!!
There are other causes for this crowding, tasting in particular, but the central bureaucracy has created the mess we have today.
Perhaps, to quote a smart Aleck twenty something year old, we should have allowed a thousand flowers to bloom.
>However, I have a question for the anarchist in Eric. Why have this centrally decided at all?
Probably because nobody could think of a way to design a DNS that forbade the same name from being (accidentally) allocated multiple times without having a root node.
Oh, I should say that of course the IETF did do something like that, however other bodies set up an actual DNS that served the TLDs, and that became the default. It was this second part that was the mistake. (You’ll excuse me if I don’t recall the agencies involved, they seem to have changed a lot over the years, and a lot of it happened before I was an adult. I don’t think ICANN existed at that time, but I am sure someone will correct me.)
This would actually be a pretty cool Point Of Divergence for an alternate history novel.
Instead of randomly dropping in on important meeting in 1983, ESR gets the flu or is distracted by some redhead or whatever
Lame Internet TLD scheme gets passed
Internet ends up being much more top down-regulated by government
There is no .com boom due to this
Chaos theory takes over at this point and the world is radically different
Gimme the Hugo in 2011, plz :p
You actually quoted Mao to make a point about freedom from centralized structure? Wow. I’m pretty impressed that you didn’t get the boot too.
> You actually quoted Mao to make a point about freedom from centralized structure?
I did not know that was a Mao quotation or approximate quotation anyway. I use that phrase all the time, and now, knowing its origin and the subsequent history behind it, it tastes very bitter in my mouth. It seems that if you were one of the 999 who disagreed with Mao, you didn’t end up blooming so much as pushing up the daisies. He really was a monstrous man.
Thanks for teaching me something though.
True, but it wasn’t seriously long after that that personal computers were connected to the Internet, or at least UUCP. I’m not sure exactly when Tim Pozar wrote UFGATE, but I can say with a good degree of confidence that you could ask John Gilmore. He’d definitely know since he helped Tim write it.
Cool story, though. And it probably is all your fault. :)
>However, I have a question for the anarchist in Eric. Why have this centrally decided at all?
While the IETF can vaguely be called “centralized”, it’s both the most open “governing” body I’ve ever heard of (contributors can come and go at will), and entirely voluntary. The IETF doesn’t tell you what you have to do, it tells you what most of the players on the Internet have decided to do together, and you’re free to go your own way if you like (for instance, by running your own naming service internally that isn’t DNS-compatible). Just look at Kerberos and Active Directory (or SIP and SIMPLE).
Because convention is not decree?
Jessica Boxer> Youâ€™ll excuse me if I donâ€™t recall the agencies involved…
I think that would be the NSF who IIRC took over the duties of administering what became ARIN in the early 80s.
And in addition to Eric’s response of the reasoning for centralized naming, I think you can add one other reason: Bandwidth. If you had to check a name an dozens (or hundreds) of locations to see if it existed, it would take considerably more bandwidth than checking just 2-3 places. In the 80’s, bandwidth was a VERY scarce commodity (my own university in ’88 upgraded from a 56K sync modem connection to THENet to a T1 at great expense), and since DNS was overhead, it was not a good idea to waste that commodity on things like name resolution. This, and the frequent existence of non-persistent connections (i.e. UUCP connections) were the primary reason that large, distributed discussion groups took place over Usenet (NNTP) and not mailing lists (SMTP).
Prior to DNS, name resolution was decentralized: there were hosts files. Surely you must be old enough to remember UUCP-style bang paths? *shudder*
Fortunately, they were going away by the time I was in my teens.
>There is no .com boom due to this
It’s funny you mention this, because a well-known tech journalist once argued in public that I caused the dot-com boom – that The Cathedral and the Bazaar was what kicked investor euphoria into high gear in 1998-1999.
Er, no. I don’t think so. If any individual was responsible, it was that fraud Bernie Ebbers with his ridiculous Internet growth estimates. He conned a lot of people into thinking any investment anywhere near the supply chain would be a magic carpet ride.
But while I think crediting/blaming me for the boom was mistaken, it wasn’t obviously crazy – not given the impact of the paper and its timing. So that, too, made an entertaining story for me to tell for years afterwards.
An idea that might help decentralization and reduce the domain name crowding problem, is as a user types a domain name, the browser could display an auto-complete list that included alternative TLD choices, sorted by order of popularity or other relevance metric. The browser could also display a small preview next to each site choice, omitting choices that point to the same site and/or have same owner. Then some TLDs could establish policies for new registrations that dictate that the owner may not be in the same, nor in the same controlled group, as the owner of the same name at any, or certain, other TLDs.
Resource reservation protocols aren’t THAT hard to design.
>Resource reservation protocols arenâ€™t THAT hard to design.
No, not if your federation of resource managers has a root. Try it without, in the presence of nodes with unpredictable uptime. Gets real nasty real fast.
(I looked at this problem twelve years ago while designing a federated indexing system for open-source packages. It was called Trove, and influenced the design of SourceForge and later forge engines.)
> If any individual was responsible, it was that fraud Bernie Ebbers with his ridiculous Internet growth estimates.
The dot com bubble was just one in a series of bubbles enabled by the Fed’s policy of holding interest rates low and decoupling them from any market feedback forces. That currency and credit is going to flow into the economy somewhere, and it will drive the prices up wherever it does.
> The dot com bubble was just one in a series of bubbles enabled by the Fedâ€™s policy of holding interest rates low and decoupling them from any > market feedback forces. That currency and credit is going to flow into the economy somewhere, and it will drive the prices up wherever it does.
Embarrassingly I formerly agreed with your thesis, although much of that may be true, I now realize Greenspan was correct, the central banks do not have control over the forces that *are still* driving interest rates lower. Essentially there is an oversupply of idle human capital, because there are vested interest barriers (include Thiel’s “China” vs. China insiders) to maximum marginal utility of production. Although the fiat (Yuan peg, gold+CPI suppression, etc) is complicit, as Thiel astutely pointed out and I re-iterated in the math of perpetual deficit expansion via halving of the fiat interest rates, fiat is increasingly irrelevant as we approach the narrow bottleneck of those who have grasp of true capital, i.e. the way to tunnel through the vested interests to germinate a billion flowers of maximum utility.
esr> they didnâ€™t kick me out of the room for being an asshole when I actually declaimed the phrase â€œLet a thousand flowers bloom!â€
I think Mr. Raymond had a profound Butterfly effect.
Peter Thiel pointed out that the extreme valuations were indicative of the narrow gate to good globalization, and I think capital was desperately seeking the internet because it was the last hope for the fiat holders to remain whole. And the smart capital that understood the economics of the internet flourished. There simply is too wide a disconnect between the nearly unlimited supply fiat+derivative capital (i.e. ignorance and lack of capital) and the true capital that knows how to maximize the evolution forward. The fiat holders have given up and are buying sovereign bonds in a death spiral of ever lower interest rates. These perpetually lower interest rates ultimately destroy ever capital intensive industry, because the insiders (oligarchy) can re-finance and eat the others. And this precisely what must happen for the billions of termites to win and evolution to march on.
I can not reasonably copy edit this post well in this small edit box given all the html links clutter I have included, so please excuse.
Eric, this is the same problem as free-market law. Same problem as multiple land ownership registries. YES, it can be done without a root. There’s always going to be a small number of root providers, with slightly different policies. They’ll have an agreed-upon policy for name reservation which nobody enforces but everybody followed because the people who don’t follow it get ignored.
Yes, there are problems. There are WORSE problems when your root is a monopoly.
>Eric, this is the same problem as free-market law.
Huh? I don’t think they’re isomorphic at all. For one thing, in the resource-manager case, you can simplify the problem to solvability if you’re willing to maintain and update a list of local roots which you consider authoritative and check with each one each time you do a transaction. The free-market law case cannot practically be solved that way because authority checks are slow and expensive.
Hrm. I’m not sure we’re talking about the same thing. In both of these systems, doubts about applicability can be resolved by reference to one authority. Trouble is that authority is a monopoly creation of the state in the case of law, or monopoly creation of ICANN, but oh, I see that I’m repeating myself. If you don’t have an authority, then you need to look at your context to see what rules apply. A law subscriber has an id card, much like a AAA card for drivers, or credit card for purchasers, or a library card for bookworms. Before a private law enforcement service can detain someone, they would need to see the id card to be able to determine if they’ve violated the law.
Perhaps competitive land title registries versus the county clerk’s office are a better example?
An interesting possibility for domain name resolution in the absence of a single root node is similar to the transaction-checking protocol used by Bitcoin (a digital cash system). Have a number of domain name registrars, each one of which can register domains. These registrars would communicate with each other and the world whenever a name is registered, and each registrar would at regular intervals publish a list of the registration transactions that they know about, each with all relevant information (domain name registered, some identification of the person who did it, timestamp, etc.), and all digitally signed by the registrar. These lists would be published and archived by each registrar, and if a conflict came up, then those involved with checking it could look at each registrar’s different lists to check the logged registration transactions and hence determine who registered a domain name first. There are probably all kinds of reasons why this would be impractical now, let alone in 1983, but it’s an interesting idea.
>There are probably all kinds of reasons why this would be impractical now, let alone in 1983, but itâ€™s an interesting idea.
It’s actually rather like the resolution protocol I came up with in 1998 for Trove. Dunno if I ever published that part, though.
Originally the domain names in Canada were managed by volunteers at the University of BC who tried to impose a geographic solution. If you were a business in Vancouver, BC they insisted on something.van.bc.ca. Only a national organization could get a .ca domain. All very logical. Businesses and individuals choose to go with anything-goes.com instead and in 2000 the Canadian Internet Registration Authority took over and new rules did away with the geographic requirement.
Russel, I understand that if authorities don’t abide by the accepted rules, i.e. if they cheat on the timestamps they claim for transactions to gain advantage, then they will be ignored (banished) by those who desire the rules. The threat to this model is that a trusted authority with a significant mass starts to cheat, but this can be prevented by a plurality of caches of transaction history. So the model boils down to any one can transaction, if they broadcast immediately. This should work well for domains since lookup is much more frequent than change. Someone could implement this for a new TLD and donate code to the browsers.
The IETF wasn’t formed until 1986, and the DNS system (with .COM, etc.) wasn’t nailed down until 1984 (in RFC920). I say this just to point out that abolishing .com, etc. doesn’t make any sense in 1983/1984, nor does being at an IETF meeting.
What you attended must have been some meeting prior to the IETF and around the time the first top-level domains were decided upon by Jon Postel etc.
>The IETF wasnâ€™t formed until 1986
Hm. I think it’s true that nobody self-identified with the acronym “IETF” at that meeting, but this was definitely the people who became the IETF whenever it formally constituted, and the group behaved like it had existed for a while – they were used to making decisions like this one, and it showed. Was it called “Network Working Group” at the time? That header, of course, was still on RFCs until at least 1996 – I had occasion to look at RFC1924 recently.
The first IETF meeting was on January 16, 1986, consisting of 21 U.S.-government-funded researchers.
You weren’t there. You might have been at a USENIX in 1983/1984, but the DNS wasn’t being designed there.
This seems a LOT like your claims of contributing to GNU prior to the GNU project being kicked off.
>We could just ask David Mills or MIke Oâ€™Dell if you were around, of course.
You could, but they’d have no reason to remember me. I was some random kid, and I’m not sure my name even registered.
>This seems a LOT like your claims of contributing to GNU prior to the GNU project being kicked off.
Quite. In both cases, I was lucky enough to be in the right place before these projects took on the organizational structure we now associate with them.
The GNU case was a bit less random; I’d known RMS personally since 1976 (before he assumed his prophet-out-of-the-desert persona) and been a contributor to Gosling Emacs (the direct predecessor of GNU Emacs) since early 1983 – actually, to be precise, I was contributing to UniPress Emacs, a productized Gosmacs variant. When Richard decided to form GNU I was one of the earlier people to hear about it. He pitched it to me personally at Boskone in…hm, it must have been either ’83 or ’84. Probably it was ’84; it was definitely before the GNU Manifesto in ’85. By the time the FSF was formally launched in 1985 I was already a GNU Emacs contributor, building on my previous Gosmacs work.
>By the time the FSF was formally launched in 1985 I was already a GNU Emacs contributor, building on my previous Gosmacs work
And here’s a curious side note. My wife, an attorney just out of law school in 1984-1985, was also friendly with RMS – she knew him from SF conventions all three of us attended, mainly Boskones. As a result, she performed one of the earliest independent legal reviews of the GPL – 1.0, not the 2.0 version everyone knows about. It’s possible her review copy (fanfold paper with notes scribbled on it in red ballpoint) is still moldering in our basement somewhere.
The pre-cursor to the IETF was GADS, not “Network Working Group”.
We could just ask David Mills or MIke O’Dell if you were around, of course.
Seriously, you weren’t there.
And yes, it was NWG: http://www.dei.isep.ipp.pt/~acc/docs/arpa–2.html
See also: http://www.byte.org/one-history-of-dns.pdf
Paul Mockapetris published RFC 882 in November 1983.
Reynolds and Postel published RFC 920 in October 1984.
Your contribution would have had to happen between these two.
>Your contribution would have had to happen between these two.
That argues for ’84…but I’m pretty sure the Usenix/Uniforum where this occurred was a summer meeting, June or July. Let me see if I can dig up a date for the ’83 meeting …. Hm, records are scanty that far back but this page of bibliography entries by date implies that both the ’82 and ’83 meetings were summer events. Furthermore, it locates the ’83 meeting in Boston and the ’84 meeting in Toronto. Both data make be pretty sure it was ’83; Boston is my hometown and familiar to me, while I would have remembered having to fly to Toronto.
This leaves us with the interesting question of why the NWG crowd was discussing TLD policy before RFC882 got written. On reflection, this may not be so mysterious. On the one hand, the DNS hierarchy evolved from naming conventions that were already in place on ARPANET. On the other hand, 882 was probably written well after the prototype DNS implementations. I see that 882 issued in November ’83; it’s quite reasonable to suppose that DNS servers were already in trial use five months earlier.
Or…I could be wrong and it was ’84 after all.
USENIX/Uniforum in 1984 was January 17-20, 1984, in Washington D.C.
> implies that both the â€˜82 and â€˜83 meetings were summer events. Furthermore, it locates the â€˜83 meeting in
> Boston and the â€˜84 meeting in Toronto. Both data make be pretty sure it was â€˜83; Boston is my hometown and
> familiar to me, while I would have remembered having to fly to Toronto.
Back then, Usenix had both Winter and Summer conferences.
Summer 1982 (not 1983) was Boston. Summer 1983 (not 1984) was Toronto. (Are you sure you went?)
Winter 1988: Dallas
Summer 1987: Phoenix
Winter 1987: Washington, D.C.
Summer 1986: Atlanta
Winter 1986: Denver
Summer 1985: Portland
Winter 1985: Dallas
Summer 1984: Salt Lake City
Winter 1984: Washington, D.C.
Summer 1983: Toronto
Winter 1983: San Diego
Summer 1982: Boston
(I had to pull the old proceedings out to get this.)
So if you attended in Boston, it was June/July 1982.
The Domain Name System was conceived in RFC (Request for Comments) 799 in 1981. RFC 799 was written by Dr. David Mills.
RFC 819, written by Jon Postel from ISI ) & Zaw-Sing Su from SRI in 1982 built on the earlier work by Mills and gave the first general outline of the DNS structure and how it would allow for easier cross-network access.
In November of 1983, Dr. Paul Mockapetris, also from ISI, published a request for comments to the Internet community entitled â€œDomain Names â€“ Concepts and Facilitiesâ€ as RFC 882 and â€œDomain Names â€“ Implementation and Specificationâ€ as RFC 883. These two papers (later made obsolete by RFC 1034 and RFC 1035) outlined a completely new way of managing host name lookups. Most importantly, they included two very important concepts, delegation and authority.
Jon Postel and Joyce Reynolds published RFC 920 in October 1984. RFC 920 is significant to this discussion: it outlined the initial top-level domain names that would be added to the DNS when it was finally deployed. These included .com, .net, .org, .edu, .gov, .mil and .arpa.
So the change happened prior to October 1984, but after November 1983. You attended USENIX in Boston, during the Summer of 1982.
Prior to the publication of RFC 883 (the Nov 1983 date), no “implementation” could exist. As I’ve shown, if you attended Usenix in Boston during the early/mid 1980s, you attended far too early (Summer 1982) to influence the decision about top-level domains.
We’ve shown that the IETF didn’t exist at the time.
I submit that, at best, you put a point forward , as you’ve said, a ‘nobody’ at the time. Others made the decisions, authored the designs and wrote the code.
Or perhaps you just weren’t there.
>You attended USENIX in Boston, during the Summer of 1982.
That’s not possible. I was definitely working for Rabbit; they paid the plane fare. Therefore it had to be after May 1983.
>You can’t have it both ways, Eric.
Either you’re confused or Usenix’s records of its own timeline aren’t consistent, or I’ve misread them. According to the proceedings page I saw, it was Boston in the summer of 1983 and Toronto in the winter of 1984 – that’s why I think it was 1983. If that’s backward – Boston was in ’84 – then the chronology gets easier to understand; that would have put me in the room with NWG right after RFC882 rather than six months before it, and after more than a year working for Rabbit rather than barely two months in.
In any case, it doesn’t matter to the story whether it was ’83 or ’84 (a possibility I allowed in the original post). The point was that the NWG/IETF/whatever displayed excellent judgment and uncommon mental flexibility by not throwing the unwashed n00b that was I out of the room, and we all benefited thereby.
I’ve said it before, I said it in the post, and I’ll say it again here: the credit for that good outcome goes mainly to the NWG/IETF guys and the culture we shared, not to me as an individual.Â So kindly stop reacting as though I’m claiming to have saved the world in a blaze of glory and need to be shot down, ‘kay?. It’s just making you look jealous, really.
“Some Other Guy”: I’m all for people criticizing blog posts where they believe them to be inaccurate, but when you come out blazing like that it would be preferable if it wasn’t from behind an anonymous account.
>when you come out blazing like that it would be preferable if it wasnâ€™t from behind an anonymous account.
The custom I’ve established here actually permits that – I bend over backwards to enable criticism, even hostile criticism, as a matter of principle. Masquerading as someone is a crash landing, anonymity is not. Zero-content flamage can trigger exile to moderation, but I don’t ban if I see even the faintest semblance of a rational argument.
Besides, do you really think this guy would be any less of a rude asshole if he were giving a name? Alas, I know better. Haters are haters.
> Thatâ€™s not possible. I was definitely working for Rabbit; they paid the plane fare. Therefore it had to be after
> May 1983.
I suggest you search for “Boston” or “Summer 1982” in the following:
The best one?
Author: Andreas Bechtolsheim
Title: The SUN Workstation
Sun was founded on February 24, 1982.
> it would be preferable if it wasnâ€™t from behind an anonymous account.
I’m sure it would, but its not going to happen for reasons I’m not going to explain here.
> Thatâ€™s not possible. I was definitely working for Rabbit; they paid the plane fare. Therefore it had to be after
> May 1983.
Rogue: Where it has been, why it was there, and why it shouldnâ€™t have been there in the first place.
M. C. Toy and K. C. R. C. Arnold.
(Ken is a friend.)
Here is another one which should hit close(r) to home:
Horton, Mark. “The New Curses and Terminfo Package”. USENIX Conference Proceedings. Summer 1982. Boston, MA: USENIX. pp. 79â€“91
> Besides, do you really think this guy would be any less of a rude asshole if he were giving a name?
I’ve seen worse from you on this very blog, Eric.
You just don’t like the fact that others here are as old and have been around the halls of *nix as long as you.
Perhaps when I’m over 50, my memory will fade, as your appears to have done.
> Besides, do you really think this guy would be any less of a rude asshole if he were giving a name? Alas, I know better. Haters are haters.
I wasn’t asking for him to be ‘less of a rude asshole’. Just that he put his name next to his statements.
Some Other Guy, is your point that Mr. Raymond did not have the discussions with the influential people who made or influenced those made the design decision?
So your assertion is that Mr. Raymond was delusional or has unintentionally rewritten history in his mind hence?
I can not detect any motivation to lie, nor can I find it plausible that his recollection could have faded to entirely inconsistent with reality. If he can barely remember many of the specifics (I can only vaguely remember 1983 and I am 45), I think it is likely than most of the participants wouldn’t remember exactly what influenced them, especially given the claim it took them an hour of group discourse to arrive at consensus agreement with the new idea.
SOG is trying to make this sound like ESR bragging about how influential he is, and wants to take him down a notch or two. I see it differently: The meeting in question happened before esr was “ESR”; he was just someone who asked a question. It’s not a testament to his incredible star power, but to the hacker ethos that a position stands or falls technically on its own merit, not on appeal to authority/prestige/etc.
It doesn’t matter if I’m a nobody; if I write some code that is demonstrably better than the maintainer of a package has blessed, people will start using my patches. At some point, the maintainer will merge the patches lest he see his userbase forked out from under him. In this case, ESR didn’t even have to demonstrate working code; he just had to point out the flaw in a geographic hierarchy for anything other than explicitly geographic entities (like the government agencies that led the early ‘net development). I’d venture the guess that 90% of the people in that room had moved across boundaries and would therefore have to change their domains under the proposed system.
I see this incident as just another example of Linus’ Law: Because ESR’s eyeballs were exposed to the proposed protocol, and the “maintainers” listened to his input, this particular bug was shallow.
Even if Eric was there, and had the influence that he claims, years later, during the early days of iPv6 and SIP, Steve Deering (who was at those early meetings) re-cast the entire thing, again, as regional.
Thr IPV6 we have now is a direct outgrowth of that.
So I too, like Some Other Guy, think esr’s claimed impact was minimal, if it actually exists.
Eric can’t even get his facts straight here. He’s like some senile grandparent, talking about the war.
Which name would you prefer he use, and what would it matter?
Either his arguments are correct and his data verifiable, or they aren’t.
Why do you feel that everyone must broadcast their True Name?
Cellphones. It used to be, on a landline, that even moving across (some) streets would mean you had to change your phone number.
Then came cellphones. You could then (in the 90s) move anywhere in your area code and keep the same number.
Now if you pick the right company you can be 1/2 the way around the world and still using the same phone number (at least for a while until your provider notices). I’ve been living away from the area code my number was issued in for almost 2 years.
There are ways to layer mobility on top of immobile technologies.
This comment is obviously written by ESR, but is under SOG’s name. Huh? Is there some bug in WordPress doing this?
>This comment is obviously written by ESR, but is under SOGâ€™s name. Huh? Is there some bug in WordPress doing this?
No, I probably botched an edit.
> example of Linusâ€™ Law: Because ESRâ€™s eyeballs were exposed to the proposed protocol
Thanks for pointing that out (it was “esr’s” not “ESR’s” :).
> years later […] Steve Deering (who was at those early meetings) re-cast the entire thing, again, as regional
How is that relevant? The .com is not regional. Without esr’s claimed input, we might have been stuck in geographic jails or perhaps even worse Butterfly effects. And how do you know Deering’s regional ideas did not originate (via the Butterfly effect) from esr’s input?
Because Deering is an actual engineer.
> Because Deering is an actual engineer.
It seems you still don’t get Linus’s law. Everyone is engineer. It is the only positive scaling law of engineering.
You may not understand well evolution and stochastic processes.
It is precisely that sort of attitude that really irks me, because it stands in the way of real progress at a time when we crucially need it. Stand aside if all you have to offer humanity is class warfare and animosity. Why can’t we work together to help each other instead?
Yeah I saw that you “schooled” ESR on the concept of Q, something which Tesla was schooling Edison and the mainstream engineeers on back in the late 1800s. No one has a monopoly on truth. In fact, you made one very big error, Tesla show that at low frequencies and high power, the bigger capital cost is in the transmitter.
I hope Eric doesn’t get banish me to moderation for this triple post (please sorry), but I forgot to add that bet it will eventually become possible to program in software the physical Q of the receiver. Perhaps it will derive from fact that we can permanently flip the poles of pairs of permanent magnets by applying a field. Flynn discovered that a pulse of an electro-magnet in between two permanent magnets, causes one of the permanent magnets (depending on the direction of the electrical current) to switch it’s dipole and become sympathetic (attached) to the other magnet, yielding not 2, but 4 times the force of one magnet alone. So it seems me we alter the physical inductance of the receiver components via software in the future.
So I think Mr. Raymond is on the correct track in focusing on programmability, that is where all the future is nanotech, biotech, etc.. The physical sciences can be reduced to logic.
> Tesla show that at low frequencies and high power, the bigger capital cost is in the transmitter.
LOL. This applies to the subject at-hand *how*, exactly?
> “everyone’s an engineer”‘
Even Barbie, apparently. I remember the dot-com days when anyone with an editor could be a web “programmer” simply by writing HTML, too. “I’m a web programmer.” roughly translated to, “last year I was a waiter.”
But real engineering? Not So Much.
Grow up, or grow a pair.
> LOL. This applies to the subject at-hand *how*, exactly?
You wrote that transmitters are always the more significant costs. Obviously it depends on what frequency band and what type of transmission. I was just busting your ego down to size a bit. Yeah I know VLF and power transmission doesn’t apply to near-distance mobile communication.
> Even Barbie, apparently. I remember the dot-com days when anyone with an editor could be a web â€œprogrammerâ€ simply by writing HTML
And if I may paraphrase from Mr. Raymond’s speech on the Linus rule, “half-assed programmers randomly banging on code can find and fix more bugs than the closed source experts”.
> But real engineering?
Apparently you don’t even know what real engineering is. I am going to teach you a lesson. Pay attention and watch the market over the next 6 months. Signing off…
> You wrote that transmitters are always the more significant costs.
Where? The receiver is always more difficult.
>> You wrote that transmitters are always the more significant costs.
> Where? The receiver is always more difficult.
Typo, I meant “you wrote that receivers are always the more significant costs”.
For Tesla’s long-range transmission of power using harmonics of the 7.5Hz resonance of the earth, the transmitter was by far more capital and more research intensive than the receiver. I covered some of the design issues at the link I provided in my prior comment.
I am not trying to say that you are not a more accomplished electrical engineer than Mr. Raymond or myself, nor am I implying the end-users will be designing software receivers any time soon, I am merely saying that Mr. Raymond acknowledged your superior knowledge in that area. I do not see the logic advantage of animosity. Mr. Raymond also suffers from overzealous religious emotional hyperbole and he censors his blog. So there is frustration and animosity. I will not argue with you on that. I am just saying why feed the disease?
I do think Mr. Raymond has made some important contributions. May he rest in peace.
I am out of here.
> And if I may paraphrase from Mr. Raymondâ€™s speech on the Linus rule, â€œhalf-assed programmers randomly banging on code can find and fix more bugs than the closed source expertsâ€.
Gives another example of the perfect being the enemy of the good.
Linux (or openbsd) isn’t perfect, but security wise is a hell of a lot better security than almost anything else out there.
Jay, you are arguing that because linux isn’t perfect then it must be worse than things you don’t like. Its a very fallacious argument.