The technology Noel is suggesting randomly chooses whether to give the current description, or some new description (which we would have to write). Then it measures which description leads more people to download Racket.
On Fri, May 6, 2011 at 10:41 AM, Matthias Felleisen <matth...@ccs.neu.edu> wrote: > > For what time period should we leave the description constant to test this > conjecture? > > > On May 6, 2011, at 10:33 AM, Noel Welsh wrote: > >> In retrospect I think this post was a bit opaque. So, some exposition: >> >> We have a hypothesis: changing the description of Racket will increase >> adoption. We can measure this and optimise for it. The measure of >> adoption could be "doesn't bounce" or "downloads Racket", for example. >> (Bouncing means leaving the page immediately. Yes these measures >> aren't perfect but the great is the enemy of the good in these >> situations.) We have various different descriptions we can try. Myna >> is a system for optimising the choice of description. A/B testing is >> the current industry standard. It is essentially hypothesis testing. >> Myna uses better mathematics to achieve better results. >> >> HTH, >> N. >> >> On Fri, May 6, 2011 at 9:22 AM, Noel Welsh <noelwe...@gmail.com> wrote: >>> On Thu, May 5, 2011 at 12:31 AM, Justin Zamora <jus...@zamora.com> wrote: >>>> A sentence like that would be a good replacement for the awful, >>>> "Racket is a programming language" currently on the front page of >>>> racket-lang.org >>> >>> We are men of science; untested hypotheses do not become us. Luckily, >>> your buddies at Untyped have recently created a system called Myna for >>> testing these kind of hypotheses: >>> >>> http://mynaapp.com/ >>> >>> We'd *love* to use Racket as a case study. I'm sure you have enough >>> traffic to get some good results fairly quickly, and this problem is a >>> straight-forward application of Myna. >>> >>> If you've heard of A/B testing, this blog post explains why Myna isn't >>> A/B testing: >>> >>> http://untyped.com/untyping/2011/02/11/stop-ab-testing-and-make-out-like-a-bandit/ >>> >>> Cheers, >>> Noel >>> >>> PS: Anyone else reading this who would like to use Myna -- drop me an >>> email at this address or n...@untyped.com. >>> >> >> _________________________________________________ >> For list-related administrative tasks: >> http://lists.racket-lang.org/listinfo/dev > > > _________________________________________________ > For list-related administrative tasks: > http://lists.racket-lang.org/listinfo/dev > -- sam th sa...@ccs.neu.edu _________________________________________________ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
