Lists Home |
Date Index |
And that Tim has considered SGML, a technology and
standard that both gave him the greatest minimal
victory of his career, and the success some hold
dear on this list not to have "mattered", is probably
the single most interesting test of his judgement
and his standards for judging.
If we had to invent XML from scratch, would it
have mattered? If HTML had been invented from
scratch, would it have succeeded?
From: Mike Champion [mailto:email@example.com]
Tim has posted his slides from a presentation at
AIIM at http://antarcti.ca/aim/Technology.ppt
Even without hearing the talk, the presentation
is pretty thought provoking. In a nutshell,
Tim looks at some technologies that did matter,
and others that (in his opinion) didn't:
?SQL/RDBMS ?Unix/C ?Open Source ?PC Client ?WWW ?Java ?XML
?OODBMS ?4GL ?AI ?VRML ?Interactive TV ?Ada ?SGML
He then goes back and rates the technologies that matter,
and didn't, on several criteria that might be expected to
?Management support ?Investor support ?Standardization process
?Technical elegance ?Apparent ROI ?80/20 point ?Compelling idea
?Happy programmers ?Good implementations ?Military backing ?
There's little obvious differentiation on most of these potential
predictors. What DOES differentiate is the 80:20 rule:
?SQL/RDBMS: 7 ?Unix/C: 8 ?Open Source: 5 ?PC Client: 8
?WWW: 10 ?Java: 7 ?XML: 9
?OODBMS: 3 ?4GL: 2 ?AI: 2 ?VRML: 0 ?iTV: 2 ?Ada: 0 ?SGML: 0
That is, all the "matters" technologies scored 5 or above (median = 8),
and all the "don't matter" technologies scored 3 or below (median = 2)
As much as this matches my own preconceptions, I'd be interested in
hearing a critique of the categorizations, rankings, and whether
looking at a different set of "which technologies matter" leads to
a different conclusion.
For example, to get the ball rolling, I can think of technologies that
"matter" mainly because of the standardization process (the US
television standard, and the GSM mobile phone standard come to mind).
I could also quibble with the categorization of 4GL as a "didn't matter"
standard -- maybe that's because dear old Natural helps pay my
salary, but mainly because I think of Visual Basic as *essentially* a
4GL even though it's not exactly what the term meant in 1980. I'd
also quibble with the low rating of 4GL on the 80:20 scale -- that's the
whole POINT of a 4GL! (Not that this would change Tim's conclusion, but
maybe others can come up with quibbles that DO change the conclusion ...)