Monday, January 19, 2004

Subconsciously, Athletes May Play Like Statisticians. The more uncertainty that people face, like wind on a tennis court, the more they make decisions based on their subconscious memory. By David Leonhardt. [New York Times: Science] The article confuses Bayesian inference at the object level — what people may do unconsciously when dealing with uncertainty — and metalevel — what scientists may do when dealing with uncertain data. It also attributes alleged prediction failures to the inference procedure, when they could come as well from poor priors. But it's still a better treatment of probable inference in the general press than I've seen in quite a while. I'll be reading the original research, it looks pretty interesting. 10:48:11 PM

I fall squarely into the draconian camp and agree with Tim Bray. Fully half of the bugs I receive in WebCore are not bugs at all, but are essentially differences in error handling and error recovery between Safari and the dominant Web browser, WinIE. None of these issues occur with XML. If we lived in a world where browsers could refuse to display malformed content (with useful error notification of course so that authors could easily repair their content), then all of these "bugs" would simply disappear.[Surfin' Safari] I understand the frustration of XML client software writers expressed in this complaint, but at another level it reminds me of the rants of grammatical prescriptivists against the “corruption” of grammar by its users. HTML and now XHTML and XML more generally are syntactic frameworks that for the most part are used to represent human communication. Is it surprising that they'll be “abused”? People with something to say tend to focus on what they want to say, not on some opaque DTDs. Furthermore, contrary to the syntax of a programming language, which for the most part has direct semantic import, a typical DTD (like XHTML 1.0 strict) has many restrictions of dubious semantic import. How many of you understand the semantic reasons for most of these claimed errors? I can't resist the impression that defenders of draconian XML parsing hold a Leibnizian hope that if only we could syntactically structure our messages correctly, their meaning would be unambiguously computable. Arithmetic (now programming language) envy is a hardy perennial in philosophies of language. The hope in Web services that the meanings of messages can be self-described is another part of this fantasy. The reality that the meaning of a message is what processors do with it is really hard for the absolutists to swallow. 10:12:51 AM