Always put bots into the same experiment

[daniel-nelson]

I see tests for not counting bots as part of the experiment results, so that is good on the experimental integrity side of things. But it appears that Vanity still randomly assigns bots to different experiments, meaning that the bots get different content—or worse, get redirected—during different crawls.

While testing in Google Webmaster Tools, Google keeps getting redirected to different homepages. . The /home/c examples redirect to /home/e, and the /home/e example redirects to /home/c.

One solution to this would be to always put bots into the first variant of any experiment.

[phillbaker]

@daniel-nelson makes sense to me. Any chance you could take a stab at a PR to do this?

[phillbaker]

Looks like this would be solved by #290.