foolip & rbyers & qyearsley - would it be useful to show these percentages to give people a better sense of the test results when scrolling through a directory, or does it conflict too much with our goal of not exposing 'gameable' metrics?

I think that's fine certainly on individual tests. You could also use a pass/total notation and make it green only if 100% to get basically the same effect. It would also show curious cases where different browsers are getting different numbers of tests, which is possible in principle at least.

I think people would get more nervous if we start to aggregate it into per-directory or total scores, so I suppose avoid that for now.

I've updated the dashboard to show results on directory pages! Check it out, here's a good example page:
https://wptdashboard.appspot.com/t/dom/events

(the full results are re-uploading now so some directories might not exist yet)

Awesomesauce! One improvement you might make is to make 99% visually quite distinct from 100%, perhaps by fading from red to white or something?

I also think it would be fine to omit the percentages, not because they are inappropriate, but they don't add much and make the results look more cluttered.

If the title could be made "Web Platform Tests Dashboard" that would also be great.

Hmm, how come only some tests say test type JS? In https://wptdashboard.appspot.com/t/dom/events I would expect all tests to be the same type.

Also, this really looks quite nice, we've all seen dashboard that are a jumble of different font sizes and weird borders, but this is just clean.

> One improvement you might make is to make 99% visually quite distinct from 100%, perhaps by fading from red to white or something?

I'm not sure I understand completely - do you mean something like this? (scale is not linear)

% Passing:
0%        50%       99%     100%
|----------|---------|-------|
100%      50%        |      green
red       red      0% red 
                  (white)

> I also think it would be fine to omit the percentages

Agreed, done

> Hmm, how come only some tests say test type JS?

This is because I only started assigning test types recently on test runs, so not all results have a test_type yet. I was planning on waiting until we did another full test run for everything to be filled in, but I can also write a backfill script to populate it now if it's confusing.

> If the title could be made "Web Platform Tests Dashboard" that would also be great.

Done

Yep, that red-white-green scale is what I had in mind. But any scheme where any failure can immediately be seen by color would do, like red-yellow-green, or just a binary red/green.