Slashdot Mirror
Why the IETF Isn't Working
An anonymous reader writes "Vidya Narayanan spent seven years working on the Internet Engineering Task Force, and was nominated for the Internet Architecture Board. But she declined the nomination and left the IETF because standards bodies are not able to keep up with the rapid pace of tech development. She says, '[W]hile the pace at which standards are written hasn't changed in many years, the pace at which the real world adopts software has become orders of magnitude faster. Standards, unfortunately, have become the playground for hashing out conflicts and carrying out silo-ed agendas and as a result, have suffered a drastic degradation. ... Running code and rough consensus, the motto of the IETF, used to be realizable at some point. Nowadays, it is as though Margaret Thatcher's words, "consensus is the lack of leadership" have come to life. In the name of consensus, we debate frivolous details forever. In the name of patents, we never finish. One recent case in point is the long and painful codec battles in the WebRTC working group.'"
with bureaucratic drudgery.
The working groups are infested by corporate types, from Cisco, Google, Microsoft, you name it. IETF was made what it was due to academics - van Jacobson, Jonathan Postel, Sally Floyd, Henning Schulzrine. No wranglings about patent rights or the need to keep their respective companies competitive edge.
We want our Wild Wild Web back, dammit!
And another thing. IT is too mature now! Let's take out all the error recovery code, and if there's an error, we'll have this routine called panic, and when it is called, the machine crashes, and you holler down the hall, "Hey, reboot it!"
[i]"...Nowadays, it is as though Margaret Thatcher's words, "consensus is the lack of leadership" have come to life..."[/i]
That can't be true! Margaret Thatcher was the most evil woman who has ever lived! NOTHING she said could possibly be true...
Really? We adopt new software every minute?
by making software patent terms shorter. Simple as that, but certainly needs legislation.
And they're just waking up to it now?
Maybe even as much as 15 years ... or more ...
You can hate on corporate types for various thing, but anyone who acts like academics know how to get anything done has never worked in academia. I work at a university and fuck me do we spend ages spinning our wheels, having meeting after endless meeting, discussing shit to death, and finally doing things 10 years after they needed to be done.
Speed is not what you find in an academic environment.
The pace of development in the last decade has been way slower than in the decade before. In 1990 there was barely any commercial Internet, no WWW, and most people were on BBSes. By 2000, we had the basis of everything which made the modern Internet. The 14 years since have been mostly about incremental improvements - tweaks and performance enhancements here and there.
As to Thatcher, she was just another front for business who did as she was told - just like actor-broadcaster Reagan. I'm not sure why those two folks are celebrated so much: they demonstrated neither principle nor originality. It scares me how much they're celebrated as figureheads of the free market when neither of them cared an ounce for it.
If you think the IETF is bad, you should look at the pace of change of warez standards. They're only reluctantly abandoning 8.3 filenames.
IETF should be written lower case, ietf. And the motto should be, "Making the internet work mostly better." After a 2-hour screaming argument with her about this, she still refused to see the wisdom of my argument.
But private sector efficiency of maximizing return on investment would also include, undermining competition by buying them out, collusion, cartel formation, lobbying the legislators, media misinformation campaigns, bribing the media personalities, intimidating critics and many other tactics. Some of it legal, some questionable, and some outright illegal.
If we confuse the private sectors definition of "maximize return on investment", even after they have openly admitted "it is the fiduciary responsibility of the directors of corporations to maximize profit", with lofty goals like job creation, low prices, wide choices, improvement in living conditions, we are the fools, shame on us, not them.
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
Take for example VHDL, a language for writing hardware.
We have VHDL '93, in the last few years synthersizes have finallly come to terms about this update. In the mean time we already have a VHDL 2008 standard. There are alsmost no simulators and definitly no synthesizers who can handle VHDL 2008. Since the first implementations of the standard are quite bad, we have decided to start writing VHDL 2008 code in 2028.
Hardware design is something open source needs to get into. However this will not happen until FPGA vendors decide to document their chips so that the open source community is able to create a place-and-route application. Sadly FPGA vendors believe that they are selling a tool chain instead of selling a FPGA. Even more sadly FPGA vendors really suck at developing tool chain, thinking that we want a graphic tool.
I think the only way out is for a chinese vendor to start selling commodity FPGAs with complete documentation, and possible start of with creating and open sourcing their place and route application. Just like how an open source UNIX system starts with a compiler, creating a healthy FPGA development environment starts with a place-and-rout tool.
I had an RFC go through a few years ago. It was an utterly trivial little thing that would have been a couple of paragraphs and maybe a week or two to get consensus in a private company setting. The RFC was about 10 pages and took over a year to get out of draft. At no point was the fundamental proposal actually objected to in any way by anyone, but little tweaks to the wornding and making certain sections more verbose. There is a lot of nitpicking in the process and a lot of discussion around mostly unimportant stuff. I'd say I had it easy having such a non-objectionable proposal to just suffer the tedium of debates about phrasing and such. Proposals which suggest anything requiring technical consensus are far more tricky.
At the same time, it feels like as of the early 2000s, the private industry has largely given up on driving improved standards in general (not just IETF, but DMTF and several other standards organizations have been relatively stagnant compared to their activity in the 90s). They've figured out it's cheaper (consensus, quicker and more profitable (patents are better than standards) to go it alone without bothering to try for a standard. Of course this leads to the opposite problem, technologies are pushed faster than they are ready. Also, it naturally creates more walled garden style experiences and less robustly federated services. For example, the big things of the 90s were email and the web. Providers were utterly interchangeable. The big things of this decade have been facebook, twitter, youtube. In the 90s, apart from cisco, network management was focused on utterly standardized mibs. Today, switch vendors emphasize proprietary interfaces that are unique for management.
XML is like violence. If it doesn't solve the problem, use more.
At one point, engineers designed and considered what they were doing. Forward compatibility was baked in from the start. Standard that don't suck ass take a little bit of time to create.
What this person wants is to call things like twitters website a standard. No planning, no design, just throw it together and keep changing it until its useful for her purpose. Where as I prefer something like SMTP, which while it evolves, it can do so in a way that doesn't require breaking changes because they put just a LITTLE bit of thought into it from the start. Okay, SMTP is a bad example for this case, maybe HTTP is better.
That leads to shit, always has, always will.
Technology doesn't change THAT fast. It really doesn't. If she actually bothered to put real thought and planning into standards, there wouldn't be a need for them to rapidly evolve. The rapid evolution of standards is because you didn't do your homework to start with and just threw some shit out there that fit what you wanted right at that exact instance. Then tomorrow ... you realize that you have a great kitchen sink, but fuck if it has a drain attached to it because you didn't bother to think about that, just that you wanted a sink.
She wants 'Agile' for standards. She can go fuck herself right in the ear. That sort of non-sense is for people who don't want to do actual standards making.
I want standards that are actual standards and aren't already 'out of date' by the time anyone knows they are a standard. Changing technology is not something exclusive to the computer age even if these people don't have any idea that technology has been changing fairly rapidly for the last 150,000 years or so.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
Tech may have developed, but has it been adapted? The world is still dragging its feet largely on IPv6. SCP, the successor to TCP, doesn't look like replacing it anytime soon. Nor will FTP or other legacy internet standards be replaced anytime soon. IETF could do better by focusing on the implementation of existing standards, rather than the definition of newer ones.
As the co-chair of the FTP Working Group & author of 3 RFCs, I can sympathize.
Coming from open source (Metalink) & working on Internet standards can be a very frustrating thing.
I pushed for improvements but made no progress.
Yea Verily.
Many many moon ago I attempted to make a comment on an IEEE standard to prevent a manufacturer from inserting a proprietary "lock". The email I received from the standard manager was to say the least, [expletive deleted].
Ever since that day I have refused to purchase any equipment from that manufacturer, and have done everything in my power to have their equipment removed from any standard 'notes and details' for equipment installation.
This is the hidden gem of "funny" on Slashdot today! Have you ever seen Chinese documentation?
When battery replace you board bang polarity over.
Yeah, good luck writing a FLOSS implementation of anything with the documentation you get from Chinese companies.
I have been working on various standards committees for as long as Vidya Narayanan. She's not wrong, but she missed one glaringly large problem:
The purpose of a standard is not to develop anything new. It is to codify common practices and references so that people can build on them. In many respects the purpose of a standard is to stop further development so that people can build things cheaply on common interfaces. If the development is ongoing (such as was the case with the CODEC wars) then it is too soon to make standards.
So, yes, it is frustrating to watch politics slow down a standards effort to a crawl. That process is called development. It is a highly political process. At some point, you have to stop development, set the standard and start building. There will be winners and losers in this process.
In other words, the ITEF is doing its job. Small groups will always be able to run rings around standards. But those small groups won't have the economic power and acceptance that a standard has.
Choose whichever side you feel is best for your products.
At least HTTP2 is going strong.
Honestly, the biggest breakage I've seen is totally a result of the lack of running-code first.
Even worse, it could be a UN meeting.
People get burned when they think of IETF as means of legitimizing industry support for their particular approach. The IETF is *NOT* that. Most RFCs turn out to be worthless summarily ignored by real world in spite of all process hoops jumped through by WG participants and reviews.
Much better outcomes are realized when IETF is viewed not as a "standards committee" rather as a service no different than github... where instead of developing your own standards process you simply use IETF leeching off existing structure, facilities, recognition, meeting spaces... while not perfect it may well be better and or cheaper than rolling your own.
This means if you want to succeed you need a working implementation first and foremost, actual users in the real world ... "working code" without interested users and or industry partners IS NOT going to cut it. Then finally go to the IETF with your I-D + LEGION of faithful consensus building followers who support your ideas.
The IETF is like a country of mostly autonomous states (WGs) ... Some WG's are oppressive dictatorships taxing oxygen you breath while others are utopias of cooperation where consensus is not merely defined by whatever the chairs want to see... Unfortunately overall governance is not all that great. One of the running jokes for me is appeals process. Having subscribed to IETF announce a millennia ago have never once read or heard of even a single appeal that was ever upheld...ever. This has grown into something of a game to be careful to check before pushing delete in the off chance hell may some day actually freeze over.
In short if you come looking for the IETF to instill legitimacy upon your idea or approach you WILL leave disappointed.
If you come to the IETF from a position of strength willing to put up with some process bullshit you stand a chance of coming out ahead.
"Intellectual Property" that is, not Internet Protocol. IETF succeeded when participants were motivated by something other than staking out as much turf to monetize. The basic premise of modern business is "do whatever it takes to get away with as much as possible", which is emphatically not part of the thought process that brought us TCP/IP, SMTP, SSH, HTTP, etc.
The problem is lawyers and MBA weasels who tell everyone that monetization is their primary duty, and that lockin and the resulting "rentier" revenue streams are the ideal course.
'Nowadays, it is as though Margaret Thatcher's words, "consensus is the lack of leadership" have come to life.'
I would never, ever trust anyone who quoted Thatcher favourably.
I keep seeing that word. Is it a new trend to use it as a verb/adjective?
FPGA vendors probably don't want to open up their specs and stuff because they are worried that opening up everything will give their competitors the secrets to what makes their FPGA "good".
Patents may come into it as well (I dont know how the patent situation is in the FPGA marketplace). And possibly a desire to stop people from being able to just buy the FPGAs at x amount per unit and force them to pay up for the toolchain too.
I've worked with a lot of engineers who believe they can understand something without practice or practical experience. Their true talent is social.
The result is focusing on the wrong things. Like the literal representation of system data instead of system behavior.
No, she is right.
The issue is not about intelligence, but Mammon. Specifically, patents.
Time and again you see great advances occur in some field right when patents expire or become un-enforceable. E.g., SSL in the '90s, 3-printing today
Nearly every organization that I have come into contact with broke down in the exact same way. A few incompetents managed to redesign the system. So it goes off the rails of whatever purpose it originally had and begins to concentrate on navel gazing. More and more is spend on things like PR, conferences, communications, legal, and most important of all, who they let in. A simple way to detect if an organization has gone rancid would be the number of MBAs who are in "leadership" positions vs people who actually know how to solve the problems at hand.
It is not so much that an organization should not have MBAs but you never give them the keys, they should be limited to marketing and maybe a little bit of accounting. But once they are in the boardroom then the organization is a walking corpse.
Another simple test is whether the original founders would even be qualified at this point to pass muster as new hires.
SCP doesn't 'replace' TCP any more than UDP does.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
QED
http://www.ietf.org/mail-archive/web/ietf/current/maillist.html
You'd think the IETF has become the Political Correctness Task Force these days. I know the main list has always been a bit of a crackpot place, but it used to be flamefests about NAT, DNS or the pros and cons of IPv6. Now it is about anti-harassment policy, diversity and all that kind of BS. They spend more time trying to protect the "rights" of trolls and the technically illiterate than trying to get some work done. It used to be that idiots were treated as such and you could tell them to go away with their stupid comments. Now, everyone should spend all their time to try to educate them and make sure they feel welcome. So no time should remain to do work as you should pamper trolls and morons.
I have 5+ RFCs with my name on them and the later ones were a total pain to get through as less and less competent people were elected to the IESG and those guys just loved to hear their own voices in every review. You got really shite COMMENTS every time a doc went to the IESG and you had to spend a lot of time to convince them that they were completely off. My favourite was this idiot in the IESG who tried to block us with "how does this relate to $TECHNOLOGY" (which happened to be his own topic). Didn't even read the fucking spec, just expected us to read $TECHNOLOGY and explain any relation to him.
I'm glad I no longer participate.
Simply stating "Corporatization" is a massive mischaracterization and oversimplification of describing the situation.
Here is a peek in history from 2005 on the IETF mailing list itself and how IETF tried to sabotage the ratification of Zeroconf (Apple's Bonjour is the best known implementation of the Zeroconf protocol). This isn't simply "Corporatization" as both Apple and Microsoft are fighting and some in the the IETF actively trying to undermine it under the guise of simply offering alternatives (that nobody wants or plans to implement and is broken by design).
Stuart Cheshire is the creator of Zeroconf and calls them out directly on the IETF mailing list in 2005.
http://www.mhonarc.org/archive...
http://www.mhonarc.org/archive...
http://www.mhonarc.org/archive...
http://www.mhonarc.org/archive...
Sarah Palin is too stupid to be really dangerous, Margaret Thatcher on the other was really dangerous because she was smart enough to lie about her evil intentions.
The biggest IETF problem I see is not enough active participation. Specifically, engineers who want the work to complete and are editing specifications, commenting based on their implementations or running working groups efficiently. Ever since the dot-com bubble burst there haven't been enough people doing that from either academic or corporate origins. Good engineers can come from either source, but unless enough engineers have the time to actually work to produce standards, the standards won't happen.
1998: "'God of the Internet' is dead "
http://news.bbc.co.uk/2/hi/sci...
"Jon Postel, a key figure in the development of the Internet from its inception, died at the weekend of heart problems aged 55."
Now, thanks to a successful internet, I have learned all about how to prevent and reverse heart disease by eating more vegetables and getting enough vitamin D (a problem for many indoors-oriented technies). Sadly, too late for Jon. Hopefully not too late for Roblimo though?
http://slashdot.org/comments.p...
The failure to adopt SQLite as a de-facto "Standard" for web browsers shows a deep problem, since a shared FOSS codebase is probably the best standard we can have.
http://programmers.stackexchan...
Contrast that with suggestions of making de-facto standards by on the ground successes with working code. Which is what SQLite has done in a whole area of embedded storage.
Like Alan Kay has said, any standard with more than three lines is ambiguous. I can agree having had to work implementing a couple standards at IBM.
A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
Require shareholder votes on all political activity by a corporation. All financial actions of a corporation should pass-through to its shareholders. Get rid of the "corporate shield".
Then shareholders will be motivated to monitor the nonsense that corporate management does in their name.
Citizen's United (and other corps) can have first amendment rights because its shareholder's speech rights pass through to it. If a corporation is partially owned by foreigners, any political action it takes should be paid for by its citizen owners only and it should be subject to penalties if it allows the foreign owners to affect its actions.
If a corporation commits a crime, the penalty for the crime should pass through to its owners!
That would be a revolution.