[racket] Computer Language Benchmark Game

> From: Eli Barzilay <eli at barzilay.org>

> Sent: Friday, February 17, 2012 8:11 PM
> Subject: Re: [racket] Computer Language Benchmark Game

-snip-
> When submissions are dropped because of a vague "it's too fast",
> that's a bias.  

For sake of argument, I'll not even bother asking you to show where anyone wrote "it's too fast", let's just look at the reasoning.

Here's a Python pidigits program that's about 50x faster than any of the other pidigits programs, and it was "dropped" immediately:

http://shootout.alioth.debian.org/u32/performance.php?test=pidigits

Of course, "it's too fast" is simply the reasonable suspicion that the Python program doesn't do what was asked (and it doesn't).


> It's a pity since the shootout started as a showcase
> of functional languages

When it was started by Doug Bagley or when it was restarted by Brent Fulgham? 

What specific things about it then make you say it started as a  showcase of functional languages?

For sake of argument, IF it was started as a showcase of functional languages would it be reasonable to suspect there was bias towards functional languages?

Perhaps your own viewpoint is not unbiased.


-snip-
> Now take a bunch of problems and throw them at a crown that tries to
> compete for speed.  In such a limited ecosystem the feedback loop is
> much shorter and the propagation of fast solutions is much more
> effective.  That makes such competitions mostly nonsensical, since
> that conceptual advantage of functional languages is practically lost.
> That's not bias, it's the nature of things.  But when such solutions
> are *disqualified* and specs change to *forbid* them, then functional
> languages lose this single advantage and get into a perpetual game of
> mimicking C solutions. 

Back in the day someone complained that it was a "brick carrying contest" and that description is very appropriate - if you don't carry the same load of bricks then you aren't even in the contest.


-snip-
> instead of FP programmers quickly coming up with new
> ways to solve problems efficiently

Where has there been any suggestion that the benchmarks game website has anything to do with coming up with new ways to solve problems efficiently!


-snip-
> All of this is bias.  (And it's the bad kind of
> bias, one where one side is completely unaware of it.  All they know
> is that "memoization" is some kind of black magic that is obviously
> cheating, and "obviously" we need to make sure that such cheating
> doesn't happen and demand that no such tricks are played.)

"brick carrying contest"


>>  Is there some way you think that differs from kindergarten
>>  name-calling?
> 
> Yes.  Please take petty flaming attempts elsewhere.

When all you write is "generally biased" all you are doing is name-calling.