Quote Originally Posted by *byronknoll* View Post 
<https://encode.su/showthread.php?p=68901#post68901>
A new rule has been added to Hutter Prize to make it easier: 
http://prize.hutter1.net/hfaq.htm#ddisc

Quote Originally Posted by *[email protected]* View Post 
<https://encode.su/showthread.php?p=69322#post69322>
It is nice that hutter prize can be sort the winner from 1 to 10 position
Think about it this way: It'd be really nice if, if it took no labor, to have 
the best score updated round the clock by tweaking parameters simply. But in 
reality, this would require the contest holder to test the same algorithm 
thousands of times. And before that happens, the programmer needs to first do 
thousands of runs. So the contest holder doesn't want to do all these tests, 
neither does a smart coder, they want big gains. However, if time elapses, we 
DO want to update the score even if it is by a small compression increase, 
because we haven't done testing for ex. a year (no winner), so we are not tired 
of doing many tests, are arms are actually waiting to do some runs on 
algorithms. If the programmer is failing to make big gains, he will probably 
want to try small gains. So I agree with the new rule. But the HP FAQ should 
explain this better. No?

Basically we want big gains, and if we actually can't, we want smaller than 
usual gains.

------------------------------------------
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T6b1c746c0845a2b9-Mfaefad8d032ca9ab04a27833
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Reply via email to