3 ms·
This kind of problems could be made public. Small or big company, government or organisation could define problem, give test input, expected output and perform
by mirekrusin 10y ago
This kind of problems could be made public.
Small or big company, government or organisation could define problem, give test input, expected output and performance points (delta size, cpu usage etc); submissions open for all.
Similar to programming competitions but instead of made-up problems, there is a real world problem to be solved.
This could be good for everybody - companies, because they'd get (I hope) the best solutions (there could be some price for top solutions, but still fraction of in-house r&d cost), for people, because they could be hired or at least would have a chance to test their skills on real problems; it could be used as part of hiring process to skip early nonsense etc.
The biggest win, IMHO, would be to have government organisations getting involved in this kind of "Open R&D" - open for contributions and also open for new, not yet articulated ideas.
- jerf 10y agoIt already is public. That's why Google is using bsdiff, an algorithm developed for FreeBSD, and which they are legally able to use due to the way in which it is licensed. We don't need some sort of central clearinghouse of problem instances. bsdiff predates the entire concept of Android by many years. To the extent that Google wasn't using it from day 1, it isn't because the solution didn't exist somewhere. (It sounds like it was due to problems that would have existed for any clever solution and they chose deliberately to ship down un-diffed-in-any-way updates.) There's actually a profound lesson about how open source really works here. And I'm not really saying this as an open source evangelist... it's more an observation that our instinctive (and probably genetic) instinct in favor of centralization is not always an accurate model of how the world either does or should work. There's not much some sort of centralized contest could do in the present that wasn't already done in the real world over a decade ago without it.