Two interesting business models for people who want to make money from open source software:
- Coverity is using its static analysis tools to look for defects in widely-used open source packages. Their current report makes for interesting reading; equally interesting is the way that some communities (such as AMANDA) have responded—their defect density dropped dramatically after stats started appearing.
- SpikeSource assembles (and sells) open source suites—Linux with Apache, PostgreSQL, Perl, etc.—that have actually been tested together, so that you can be confident they’ll work.
The common element here is quality: people will always pay for it, just as they will for medicine and chocolate. Makes me wonder whether you could create a business selling insurance on software: “You give us $199/month, and if this crashes more than once per hundred thousand hours of operation, or if someone gets through the security, we’ll pay you $10,000.” We have a lot of data on how flaky various platforms are, and we can guesstimate the cost of downtime. If any grad students in economics are looking for thesis topics…
Zipf’s Law says that frequency is inversely proportional to rank, i.e., the second most common word in a large body of text will occur half as many times as the most common. I have observed an even steeper curve for Software Carpentry feedback: of the 336 corrections I’ve received, 212 are from one person (Adam Goucher), 21 from Matthew Moelter, 12 from the next two people, and then we’re down into the curve’s long tail. Has anyone ever done similar stats on the volume or frequency of contributions to software projects?
Yes, it’s back: Google’s Summer of Code is going to give students grants to work full-time on open source projects this summer. You can start applying on May 1. Two U of T students got grants last year; I hope to see at least as many get them this time ’round.
Leah Bobet (a Toronto-area writer) has posted a lengthy summary of the Mark Tushingham case. For non-Canadian readers, he’s a scientist with Environment Canada who was ordered by our new Conservative government not to speak about the science behind his new science fiction novel.
Please, contact your MP and gripe about this: we are only free to speak if we are all free to speak.
Brian Hayes’ column in American Scientist is almost always my favorite part of the magazine. He now has a blog at http://bit-player.org/, which I recommend highly. I don’t know anyone else who writes popular computer science this well, at this level—if you do, I’d be grateful for a pointer.
The final standings in the 2006 ACM-ICPC, held in San Antonio, Texas, have been announced:
|1||Saratov State University||Russia||6||917|
|2||Jagiellonian University – Krakow||Poland||6||1258|
|3||Altai State Technical University||Russia||5||681|
|4||University of Twente||Netherlands||5||744|
|5||Shanghai Jiao Tong University||China||5||766|
|6||St. Petersburg State University||Russia||5||815|
|8||Massachusetts Institute of Technology||United States||5||831|
|9||Moscow State University||Russia||5||870|
|10||Ufa State Technical University of Aviation||Russia||5||980|
|11||University of Alberta||Canada||4||479|
|12||University of Waterloo||Canada||4||636|
You can take a look at the problem set if you want; I’m more interested in the fact that Russian schools captured five of the top twelve spots, and that there were more Canadian schools on the list than American…
Discussion about Python 3000 is heating
up. What I haven’t seen so far is a list of things that will be dropped from the language to make room for new ideas. Back quotes? Don’t think anyone will miss those. Tuples? If the language acquires a generic way to ‘const’ an object, tuples (and frozen sets) could be dropped (they’d be const lists and const sets respectively, and we could even start using const dicts as dict keys).
What else? What would you like to see removed from the language?