May 28, 2009

Structural Bias -- Standards and elsewhere

How Power Corrupts: Indirect Force

Consider the following:  if you take white noise -- sound at all frequencies -- and filter it, you get bias: if you filter out all the low frequencies, you'll get noise that sounds high-pitched. If you filter all the high frequencies, you get a low-pitched sound. If you filter everything below 339 and above 441, you will hear a 440 A tone, even though you started with white noise.

Something similar happens with software. There was a time when Microsoft was accused of introducing OS bugs that impacted other people's applications more than theirs! Of course, they hotly denied this. They hate bugs! They spend an enormous amount of energy on reducing bugs.

However, bugs in its software are "noise": fallout from adding features, working on tight schedules without much time to think.  Software vendors rely on testing. Testing is a kind of filtering for bugs.  And of course, filtering is, unfortunately, selective. It is somehow more important to eliminating bugs that affect you or your friends more than it is those that affect your competitors. The result is bias with no paper-trail: you did nothing wrong, every act is above-board, open, transparent, improving software, etc.

The same thing can happen in standards! People take positions, have opinions! There are lots of ideas generated, some good, some bad, suggestions, positions, what have you. Many contributors are independents, students, and many -- most really -- are sincere employees and software developers.

However: not everyone gets funded. Not everyone has a PR person assigned to report "this week". Not everyone is friended or employed or encouraged, or have the luxury of spending half- or full-time on their life's work.

Those with an agenda, implicit or not, apply filters: fund those whose work supports their agenda, not fund those whose work is counter. Individuals who are promoting a position which is in favor will be amplified, invited to meetings, recommended, and those whose position is not in favor will not. It's natural.  Bias is endemic. No individual needs to work according to an agenda to be a player.

Of course, I'm part of this game. I work for Adobe. I don't think I shill for Adobe, and most of my view of "the Web" was formed when I worked for Xerox, or later AT&T. But I can see Adobe's point of view, the sincerity of the people who work on its software, the lunacy of some of the silly anti-Flash attitudes not based on anything other than polemics that evaporate on examination.

Does this make me biased? Not any more than anyone else in the process.  If there's bias, you've already shown it by choosing to read this.

May 19, 2009

W3C TAG blog entry on defining language semantics

I'm still working out where/how to blog. I took a mailing list posting on 'meaning of names and operations of services' with a discussion around it and posted it to the W3C Technical Architecture Group blog. Is this useful? More likely to survive as a meme than just leaving it on the mailing list? I think this is an area of language design policy that's worth elaborating, but is blogging about it effective?


May 7, 2009

ICANN and new TLDs

Recently, on a short flight, I sat next to Doug Brent of ICANN, and we discussed the proposal for new top level domains.   I'm concerned about a couple of topics that don't really seem to be resolved:
  • Internationalization conflicts between IRI and the new (non-ASCII) domain names. As far as I can tell, the conflicting methods of encoding and display haven't been resolved. This seems like a major disaster.
  • Distributed responsibility means more changes for messing up: the more registrars and domains and domain policies, the more likely it is that someone will allow spoofing domain names to be registered.
  • Trademark issues: the assurance that the concerns of trademark holders about squatting, monitoring, and having to spend much more to protect trademarks -- well, the assurances aren't particularly convincing.
Anyway, I'm not convinced. Domain names are an area where stability is more important than "first principle answers". Yes, there's no particularly good reason why you can't have something.coke instead of, but for better or worse, don't change it unless there's a better reason for changing it than "why not?".

Living in Web 1.0

I've been on the web since near the beginning, and I've had a web page or web site since, oh, around 1996. I've only been using since perhaps 2001. But I feel stuck in Web 1.0. If I have something to say, send it to a mailing list (how archaic!), and less is more. It's time to update my communication style, and blogging more often is part of that. I expect to go back and forth a bit until things become clearer, between here (blogger), an Adobe blog (on, my own web site, starting a group web site and blog about HTML issues, and using W3C TAG blog. We'll see.

Restoring Medley Interlisp: running well on modern systems

at and on GitHub (and see issue list ) It's great to work with old friends, as if 30 years hadn't passed.  We have a...