Has two-way sync increased contributions? Yes!

[Philip Jägenstedt]

Summary: Chromium contributions to wpt have increased 220%, compared to an overall increase of wpt contributions of 54%.

Details:

Reading "Two-way sync has lead to significantly more test contributions from Chrome and Mozilla" today and really wanting it to be true, I did some digging.

Since the first Chromium export was in October 2016, I compared the periods January-September in 2016 and 2017 by looking at the output of these two commands:

git log --no-merges --since=2016-01-01T00:00Z --until=2016-10-01T00:00Z --pretty=%ae (1946 commits)

git log --no-merges --since=2017-01-01T00:00Z --until=2017-10-01T00:00Z --pretty=%ae (2988 commits, a 54% increase)

In that output, there are 561 unique author emails, and from there tried to work out the subset that could be called Chromium contributors. First, all @chromium.org, @google.com, @intel.com, @opera.com, @samsung.com. (Not @igalia.com, as some may contribute to WebKit also.) Then, I added a handful of @users.noreply.github.com and personal emails I recognize as being the same as one of the above. I ended up with 183 unique emails.

Based on this list, I found 245 (13%) commits in the 2016 period and 783 (26%) in the 2017 period. Per Ecosystem Infra Stats, there were 586 Chromium exports in the 2017 period. So, Chromium contributions have more than tripled, and it seems very likely that automatic export of changes has something to do with it.

(Excluding personal emails to eliminate my recognition bias results in 180/718 instead, which is an even bigger difference.)

I can't do the exact same thing for any other engine since I'd be much worse at recognizing emails, but the results for just @mozilla.com are 177/181.