Hacker News Reader

Comment by yongjik | Hacker News Reader

Would the author be happier if Chrome/Chromium didn't do these "field trials", and instead just enabled whatever feature they were testing for all users, and the author was hit with the same problem, along with everyone else?

When you cut through the rhetorics, the author is basically complaining that Chromium decided to update itself. Well, maybe it warrants complaint, but I fail to see what's such a big deal. (The alternative is millions using browsers ridden with last year's vulnerabilities.)

View on HN

>the author is basically complaining that Chromium decided to update itself. Well, maybe it warrants complaint, but I fail to see what's such a big deal. (The alternative is millions using browsers ridden with last year's vulnerabilities.)

You're misrepresenting mandatory enrolment in A/B testing with plain software updates. Firefox is an example of a browser that does software updates, but does not force you to use potentially buggy features - and yet they do not have "millions using browsers ridden with last year's vulnerabilities", because their alternative is to give users a choice to enroll in experiments. I do participate in experiments since I wish to improve Firefox, but I could switch to any time to no experiments at all.

A/B testing is not some nefarious scheme to harm users. When Google or Mozilla wants to add a new feature, there are basically three choices:

(1) Opt-in: Add the feature to the new version. Users must choose to upgrade.

(2) Opt-out: Roll out the new version automatically, unless the user actively choose not to.

(2a) Opt-out with A/B test: Roll out the new version to a small portion of users first, and proceed if nobody catches fire.

Once you decided to use opt-out, there aren't that much difference between (2) and (2a). The major difference is that with (2a), a smaller number of users are hit when an unforeseen bug inevitably finds itself into release process.

And of course we know what happens when something like a web browser relies on the users to actively click "Upgrade". I'm sure there are some people still using IE6.

All features are potentially buggy. There's no particular reason to allow users to opt out from trials, same as there's no reason to allow opt out from upgrades. The chance that you will be impacted negatively is minimal.

> the author is basically complaining that Chromium decided to update itself.

Exactly. Apps shouldn't do that on Linux.

Reply on news.ycombinator.com