Wikipedia talk:Historical archive/Explore

Special:Randompage isn't random though, is it?? A few pages keep turning up, eg. Netlist, Royalston, Massachusetts. -- Sam


 * I think it's explained somewhere (don't ask me where) that it works by selecting a batch of 1000 (or some number) pages, and then dishes those out at random, and after 1000 (or whatever it is) requests for a random page, it selects a different thousand. So it's not fantastically random, but it's... well, it's more random than recent changes, I guess. --Camembert


 * I was wanting to ask about the random page, it seems influenced by what you do. Recently I've edited a few football articles and on random page a lot of football articles are popping up... --Josquius 13:46, 5 Dec 2004 (UTC)

How random is the random page feature? I mean over the past few days I seems to be getting some pages more than once, and it isn't just one page. Is it just me, or is it not really that random? Just wonderimg... Smelialichu 19:43 Jan 13, 2003 (UTC)

There are so many different methods that could be used. True randomness is hard; but, pseudo randomness is indistinguishable to a casual observerer. It could be that you are only seeing a pattern in the flux. When I was a child I saw my brother roll 7's with two die an amazing 17 times in a row. Then chance moved back to equilibrium. Someboy told me that the casino record for a consecutive run of one colour is in the high twenties. If you never got a duplicate or a quick revisit to an article, that would that be weirder. Two16


 * After having tried the random page feature myself, and having some practical experience with computer pseudorandom number generators, I believe that either it's a very poor PRNG or there is some other problem causing less randomness than what one might expect. --Robert Merkel


 * It seems that way to me too. I guess I'd need to actually do a proper scientific test to be sure, but I seems to get repeat articles very often. Cuold any devs comment on this?

The random page function (see the source) keeps a queue of 1000 randomly selected pages (no duplicates), which queue is regenerated at semirandom intervals (avg time between refills should be about every 450 loads). On each load, one of those 1000 pages is selected randomly. This can produce duplicates on subsequent or relatively close loads, and with a field of only 1000 is fairly likely to if you push the button frequently. Possibly better would be to remove each one from the queue as it is loaded, then refill the queue when it's found to be empty. (Which would have the additional benefit of semiautomatically replenishing the random queue of a very small wiki.) --Brion 02:02 Jan 18, 2003 (UTC)

I had a feeling it may work something like that. I guess that's the reason for the duplicates. Thankyou for clearing that up, it's much appreciated. Smelialichu 09:16 Jan 18, 2003 (UTC)


 * I've made the above-mentioned changes to the random page selection; that is, preventing short-term duplicate selections. --Brion 06:43 Jan 24, 2003 (UTC)

Rambot and random pages -> User talk:Rambot/Random page

For random page, why not use mersenne twister? I hear it works fast, and when there is little traffic, you could reseed the generator just to keep us from predicting what will come next.

Wikipedia link
Why won't the wikipedia link (in the first line) work? I've just done a minor edit and in the preview the link works, but after it's save it's still broken. -- bruzie 23:22, 1 Jul 2004 (UTC)