Re:Why Hacker News should use a different Deterministic Algo...
by Udo on Monday December 02, 2013 @11:40AM
(#6834679)
At a certain point this distinction becomes just philosophical wankery. Make no mistake, the proposed deterministic algorithm is a random one in disguise. And that's fine. Over large sample sizes, this randomness becomes sufficiently deterministic. Which is what the original proposal suggested, and it's also why sometimes a random variable can be elegant, too.
[ Parent
| on HN ]
Re:Why Hacker News should use a different Deterministic Algo...
by larksimian on Monday December 02, 2013 @11:52AM
(#6834758)
The article's philosophical wankery was way too pragmatic.
Random vs deterministic is a spectrum, anyway. Given enough information about the state of the machine doing the random picking/number generation all computer algorithms are deterministic anyway.
Heck, given enough information about the state of the universe everything is theoretically predictable, some predictions are just really hard.
Hopefully this achieves the needed philosophical wankery quota.
Re:Why Hacker News should use a different Deterministic Algo...
by Udo on Monday December 02, 2013 @11:57AM
(#6834789)
That's sort of what I was talking about. Sorry if that wasn't clear. Anyway, I'm with you on that. :)
Re:Why Hacker News should use a different Deterministic Algo...
by kens on Monday December 02, 2013 @01:34PM
(#6835471)
The article describes an algorithm that randomly shows articles based on a computed distribution, and calls that "deterministic". That doesn't sound deterministic to me - is the article misusing the term, or am I misunderstanding "deterministic"? (genuine question)
Re:Why Hacker News should use a different Deterministic Algo...
by yummyfajitas on Monday December 02, 2013 @01:37PM
(#6835491)
The algorithm deterministically computes how many people should be shown article A and B and randomly assigns people to those different groups.
I suppose any algorithm involving sampling will be random in this manner, but the point is that randomness is not being introduced to work around flaws in the decision procedure.