(Resolved) minor bug with random page [#1505]
: (Resolved) minor bug with random page
(2008-08-02 15:24) [#4936]
As a frequent user of the random page function, it occurs to me that
- it occasionally returns Homepages, despite these being excluded. I suspect this is because it really returns the alias, which is another page type than Homepage.
- it has a tendency to pick the same page twice in a short span, against not only the popular odds but also the statistical odds. This is not even disturbing for its most frequent user, only peculiar.
: not a bug, pure math
(2008-08-02 15:18) [#4940]
regarding point 1: true. You just exclude homepages, but not aliases. As there are over 120 aliases of homepages it is not surprising that you land on a alias for a homepage occasionally.
regarding point 2: some math shows that the probability of hitting on the same page twice is about 30% if you use random page 100 times. Add in the fact that SL currently has over 2000 aliases and the odds that you hit the same page twice increases dramatically.
I have added some functionality for SL's #1 RandomPageJunkie :o)
: Re: not a bug, pure math
(2008-08-02 19:10) [#4944]
(sad face for being accused of falling for the birthday paradox) I believe it is a random function. My math did not take the aliases into account. Even then I had some strong incidences, but then again multiple experiments with the same odds will also produce anomalies ...
: Re: not a bug, pure math (update)
(2008-08-02 21:13) [#4946]
Didn't want to accuse you of anything. Actually, I'm using MySQL's rand() function about which its documentation says:
RAND() is not meant to be a perfect random generator, ...
I assume this means that rand() is good enough, but that it's not a cryptographically secure random generator.
I ran some tests and with aliases there is a 10% chance that you hit the same page within 50 tries, and a 2% chance that you hit the same page within 25 tries.
Maybe your memory is playing a trick on you: I looked through the logs and you had only three duplicates within the last three days (250 random pages served). Note: this does not take aliases into account, so you may have seen more duplicates.
: Number of pages
(2008-08-02 18:19) [#4943]
I assume that the current number of pages as listed by %% Frontpageblurb %% includes the number of alias pages. It seems more informative to exclude the alias pages from this count, no?
: Re: Number of pages
(2008-08-02 20:51) [#4947]
Probably true. One could also argue that homepages should be excluded.
Current counts are:
- Aliases: 2190
- Homepages: 1732
- Other: 13281
: ((no subject))
(2008-08-02 21:12) [#4948]
I should show this to the people who admire me for not having a TV.
I hope you paint the smilies wherever I don't bother to.