Slashdot Mirror
Bell's Axioms on Standards
CowboyRobot writes "The inimitable Gordon Bell
has a 5-page article at ACM Queue on the difficulties and necessities of standards adoption, in which he lays out 11 axioms on how to best establish and adopt standards. He quotes MIT's Dave Clark, 'Standards setting sits in a boring trough between two exciting peaks. The first is the peak of technical innovation, and the second is the billion-dollar investment or market.' It's often the rush for money that makes us move to fast to do the job right."
I completely agree. Microsoft attempts to move too fast in completing their products, and often horribly botch them. They also have been known to "Embrace and Extend" someone else's product or standard, marring it in the process, and completely ignoring the concept of standards in the first place; standards in the technology world mostly exist to make things compatible, Microsoft's "embrace and extend" usually ends up with an incompatible standard or protocol.
Patrick "Diablo-D3" McFarland || http://AdTerrasPerAspera.com
The best thing about standards is that there are so many to choose from.
6. Consumer investments are never to be undone by a standard.
I think this is the most important one - simply because if it does not make business and market sense, a standard is useless. Being an academic in the industry, I very often come across this - people in the academia want to implement standards without realizing how it would affect the market, and the impacts that it would have on existing businesses. And then they wonder why the standards failed.
Why? Because unless it is beneficial to the businesses at large, and makes technical _and_ business sense, standards are not going to be implemented. Not only that, changing or implementing a standard is going to break a lot of existing stuff, and companies often need to see some kind of adoption before they jump in.
And often, if a few big businesses decided not to jump in, it is going to fail because they would control a large chunk of the market. This is not always true, but sometimes it affects how things work - classic example being that of IE. IE does not follow standards, yet it is popular. Webdesigners design to make things work in the real world (tm) browsers, and not in compliance to XHTML standards. Why? Because it makes practical sense.
Ofcourse, sad part is that business often do not respect standards, and those setting standards do not sufficiently respect business practices. Classic Catch 22.
standards setting sits in a boring trough between two exciting peaks. ... It's often the rush for money that makes us move to fast to do the job right.
Standards (now) exist in spelling, ever since Dr Johnson's dictionary came out. Before then, spelling was whatever someone could be bothered with.
Or did I misinterpret the last sentence, and what was actually meant was that the rush for money makes us abstain from food to do the job right?
"She's furniture with a pulse"
Well, in all fairness, standards are useless if nobody follows them.
If Microsoft can design standards that the rest of the industry would follow, I'd like to see them do it.
Really, it's quite irritating when you have a standard that the big guys don't follow - you usually design systems which work for others, and which works for the big guys - almost two or more standards.
If MS says they are going to design standards, I suggest we tell them they can - but on the condition that they should adhere to what they design, and not go around breaking it all the place. Maybe tie it up with fair-trade practices and stuff, just to make sure.
That way, we could atleast -know- what the hell is it that they are breaking, and design our systems to accomodate that.
But I don't see that happening because unlike most other big businesses, Microsoft tends to be very -petty- and -cheap- when it comes to following standards. Which is what sucks.
It's one thing not to follow standards for business sense, like I mentioned elsewhere. It's another to do so just because you're a powerful behemoth.
Take even the simple case of support for PNG - why the hell can't they go ahead and do it? Oh wait, it's not because of business or the market. It's because we're fucking big and don't care two hoots about you or your petty little problems. It's that kinda attitude that _really_ pisses me off.
on achieving standards compliance when his own site doesn't validate at all.
s in action.
As I've said elsewhere, my original trade was engineer, so I'll use an engineering standard that you will ALL have handled, and hopefully it will be distant enough from computing to illustrate some points without getting embroiled in a fanboy flame war.
Lots of people say the original great invention was the wheel, as an engineer I'd have to pick the much more recent lathe bed or screw thread, without the lathe bed NO accurate machining is possible (including screw threads)
What this means in laymans terms is that EVERY SINGLE COMPONENT would be bespoke, hand made, and slightly different, so EVERY mechanical device would have to be ENTIRELY hand made... no going to the shop to buy a gasket, you have to make every one by hand to match whatever you are working on.
When we come to screw threads there are many standards, BSP, JIC, Metric, and variations amongst these.
Each thread has it's own strengths and weaknesses, and included in these is the actual cost of making a given nut or bolt to a certain thread patterm, which depends on the specification and STANDARD of the thread chosen.
BSP is simply the best for hydraulics for example, while BSF and UNC are good for bolting things together, however metric comes along, with it's predisposition (which is an entirely ETHICAL predisposition, eg academic, rather than practical, and likes everything to be in nice multiples of ten) for simplicity, and it has to be said this simplicity makes a metric bolt cheaper to make than say a BSF bolt. (if you think the cost is the same in these days of CNC then you are wrong and you are omitting precision tolerances and cost of metal removed etc etc etc) so of course metric bolts rapidly became extremely popular.
Trouble is metric threads are, to an engineer, shit. They are cheaper, so management selects them over the engineers preference.
Henry Ford went metric on the truck side of things many mnay years ago now, but for many years (and possibly still today, I'm out of the loop) kept some things, notably wheel nuts, on an imperial thread. Why? because the metric threads worked loose.. like I said, metric threads are crap, old imperial threads are far superior.
SO, to the point.
Obviously a standard thread is very useful, a 1/2 inch AF spanner bought in the UK will fit perfectly around the head of a 5/16 bolt made in germany, which will screw perfectly into a 5/16 water pump made in the USA, because of standards.
SO, we have a general trend away from the old, superior, but more expensive, standards, towards the new, inferior but cheaper standards.
This is all great in theory and the halls of academia but in the real world these unequal pressures bring about strange offspring.
About 20 years ago I worked in Spain, which had just thrown off the mantle of a dictatorship, and was working its way towards joining the European Union, so things including the economy were in a flux, although spain was metric lots of the stuff that was used in spain (for example austin had a car assembly plant there) was not metric.
Instead of everyone owning a whitworth tool set, an AF tool set, and so on (remember that thing about costs, imperial spanners were higher precsion than metric too, so cost more) so the spanish came out with a unique solution, which saved EVERYONE money.
When they made a bolt they would make a 5/16 UNC bolt, just like everyone else, except it would end in a METRIC head, so pable with his metric toolkit could work on it.
When you come across this and see a "wrong" size bolt head on say an engine, you ASSUME that some bastards have ruined it in the past and stripped the thread, and drilled to a larger diameter and tapped it to a matric thread, which of course will have an utterly different ideal torque setting, so you sigh and swear.
It is only when you REMOVE the bastard that you discover that it is a 5/16 UNC.
At this point, all the preceding standards are vastly degraded, you cannot as
http://slashdot.org/~GuyFawkes/journal
ACM Queue vol. 2, no. 6 - September 2004
by GORDON BELL, MICROSOFT BAY AREA RESEARCH CENTER
After reading this Microsoft employee's views on standards, I think it's time to rename the place he works to
"Bay Area Research Facility".
I think it is a frightening thing, but very much worth noting, that since the Internet left academia in the early ninetees, not a single new protocol or application has a widely accepted standard. Every single new application that has come along since then has been a hodge-podge of incompatible solutions.
Case in point:
IM - ICQ came first with a proprietary protocol, similiar applications were made AOL, Yahoo, Microsoft all using incompatible, opaque protocols. Industry talk of standardising is a joke, and free software standards like Jabber have no market share to speak of.
Audio: Real, WMV, AAC, etc etc all fighting for acceptance. The only existing standard is MP3, but that is only because it was used by hobyist copiers and never got any corporate attention until it was entrenched.
Video: How many times have you seen this: "Please choose video format: Real - Quicktime - Windows Media". I wonder how much it costs web publishers extra to encode everything in three different codecs, and when the end result is still that you are tied to proprietary players in Windows (and maybe Mac).
Voip: At least here there is some effort, and we have a whole host of different standards, H323 and SIP etc etc. But mostly different companies services are incompatible, and most users use proprietary game chatting software, or Skype.
P2P: Lots of different vendors developing incompatible programs with as opaque and complicated protocols as possible.
Vector animations: Flash...
The only applications for which we have standardized protocols, email, the web, ftp, etc, are those that were around before the Internet became mainstream. I cannot think of a single credible counterexample. I think that is pretty safe to say that Internet standardization is not rare or difficult: it is dead!
Not sure which I am but to make my comment, I must betray a bit of both.
When I got my first software engineer job, a PDP-20 was a cool toy. Back in the day, one thing about DEC that just blew us away was that they actually published the bus standard. Any garage or EE grad could design peripherals to plug into a PDP backplane. It was a defacto standard and it turned a "good" processor family into a great product family because any would-be customer in a laboratory could just about be certain that whatever odd data-capture, storage, display or comms device was needed, someone had developed. It made DEC's fortunes to multiply the advantages of a [then] cheap new processor by the advantages of a published "standard" architecture. The ironic thing is that when the PC revolution got under way, DEC's response was to emphasize the PRO350, a [by then] relatively closed PDP based architecture over its "Rainbow" product. The IBM PC architecture rules to this day because IBM beat DEC at its own game. Those of us in the early 70's who said "Digital is crazy: they're gonna let everyone else make the peripherals that DEC could be sellng" just didn't get it. Gordon Bell was a chief architect and engineering manager of the original PDP...he did get it, at least back then.
There are lessons in the parabola of DEC's fortunes that bear study by any of us, engineers to marketing, who compete in development of technical products.
Favorite geek-geezer link is to a book that recounts the history on which my pathetic career was spent. See also a link to much of Bell's commentary and output since joining Microsoft
SLASHDOT: news for people who can't concentrate on work or have no life at all and got tired of yelling back at the TV.
(T)he International Standards Organization (ISO) and the International Electrotechnical Commission (IEC) designated Oct. 14 as World Standards Day to recognize those volunteers who have worked hard to define international standards ... The United States celebrated World
Standards Day on Oct. 11; Finland celebrated on Oct. 13; and Italy
celebrated on Oct. 18. -- Open Systems Today, 10/31/94
in other comments here I have mentioned a bit of history demonstrating how standards can foster the success of your company. But it occurs to me that the situation is more pervasive and more universal when it comes to the benefits of standards to folks who develop software.
The essence of the situation is that virtually all of us who develop products or systems are integrators. No matter how much original code we concoct, we no longer write whole systems from scratch. Instead, we succeed or fail according to how well our bit plays with all the other bits that combine to function in the way the users perceive. Any grown-up software engineer simply MUST either look deeply into the work that [many] others have done or trust and understand the standards to which those others have presumably worked.
SLASHDOT: news for people who can't concentrate on work or have no life at all and got tired of yelling back at the TV.