Chromium web platform contributor values survey

56 views
Skip to first unread message

Rick Byers

unread,
Nov 14, 2019, 1:17:56 PM11/14/19
to blink-dev
At BlinkOn this morning I asked attendees to fill in a values survey to try to help us all better understand what principles matter most to the community in their work on chromium. In particular, what are the things which we mostly all agree are important to chromium, and what are the things where we have unique perspectives? As a community we often need to talk about and agree on fundamental tradeoffs here (eg. between maximizing engineering velocity and minimizing process overhead).

If you contribute to the web platform portions of chromium, please consider filling this in. I'll be presenting initial results at the end of BlinkOn tomorrow, but we'll leave the survey open and share the results more broadly.

Thanks,
   Rick

Rick Byers

unread,
Feb 28, 2020, 9:51:32 AM2/28/20
to blink-dev, Ashley Haman
I just realized I failed to follow up on this, sorry*

We had 39 responses to this survey during BlinkOn and another 10 afterwards. That's a small sample of our community, but I think there are patterns in the responses that are large enough to be significant. Here are an updated version (to include all 49 responses) of the slides I presented at the end of BlinkOn with my key take-aways (from survey analyticsraw data):

image.png

image.png

I've pointed at these results a few times in various discussions when arguing for the importance of some principle (eg. Ensuring people feel respected in our communications), so I've already gotten a lot of value out of it. Thank you to everyone who contributed!

Anyone else have any thoughts on the implications of this data for our community, or what we should do with this sort of thing in the future? One person reached out to me asking that we also try to take a pulse on the community on how we feel we're doing against these values. Thoughts? Should we consider doing some larger chromium open source contributor survey? The relatively low response rate for this survey makes me think we probably shouldn't (since we'd always question whether we're looking at a very biased sample of the community). 

Thanks,
   Rick

*I had trouble getting the results public in a nice way without sharing e-mail addresses, but I finally just deleted all the e-mail addresses and that worked. 
Reply all
Reply to author
Forward
0 new messages