On Thu, 2008-10-09 at 15:47 -0700, Isaac Gouy wrote: > > Exactly. That is what I propose, not just a contest between > > language > > implementation but between language advocates. The shootout works > > that > > way anyway, people constantly tweaking their language or their > > implementation. > > > > No, the benchmarks game does not work that way. > > FAQ > http://shootout.alioth.debian.org/u32q/faq.php#contest >
Are you referring to the requirement that that "same algorithm" be used? I think what I would personally like to see is as much leeway as possible so that you can explore the full potential of your language, which to me is much more fair that specifying EXACTLY how each line of code should be written. So I think it's fair to simply specify what is required to be functionally identical to a reference implementation. You can even cheat if you can figure out how to. To be sure, you would have to be able to pass a stringent battery of tests. If it cannot be proved that your black box is different from the reference black box, then your implementation is legal. We have noticed that even minor eye rule changes are easily detected by the output statistics. Of course the source code is always available and others can steal any good ideas. Also, if someone finds a clever cheat it should be possible to create a test to reveal it. - Don
signature.asc
Description: This is a digitally signed message part
_______________________________________________ computer-go mailing list [email protected] http://www.computer-go.org/mailman/listinfo/computer-go/
