Applying the "make it an option" tactic to all questions of UI is how one ends up with a preference menu like Eclipse's. It's a tradeoff, and "I'm on the losing end of the tradeoff" doesn't mean the wrong decision was made.

I think that the second sentence mischaracterises my argument (I haven't said anything, though my opinion is probably clear, about whether this is the right or wrong decision); but I appreciate the first sentence. I have been arguing the drawbacks of one extreme of the trade-off (bundle everything!), but it's worthwhile to observe, as you did, that the other extreme (prompt for everything!) is equally silly—so that really, I think, what we're all debating is what is the appropriate point in the middle, not at which end we should live.

The button is "decide on first click" whether you want to actually set up the integration. I think the opting aspect is fine, even if it should have been constructed as an extension.

That does not address the terms of service concerns. Not to mention that that sort of thing (random popups asking "do you want to enable X") gets far too many accidental misclicks. Look at bundled installers to see what I mean.

I'm afraid not, because for the average user, who understands very little about software, asking them "do you want [complex new feature]?" is almost always going to get a "no". And even if it isn't an actually complex new feature, it sounds that way to them.

It's true that this seems perhaps a little silly to people like us. We might prefer this to be opt-in. But we are a rare type of user. Studies show the best decision here is to bundle it by default.

> Studies show the best decision here is to bundle it by default.

I believe that there are meanings of 'best' for which this is true, but they aren't, I think, necessarily the right ones. In this case, it seems like 'best' is "best for widespread adoption of Pocket".

OK, so let's buy that widespread adoption of Pocket is a good goal because more users like it than don't. I'm sure there are lots of add-ons with this property, but we don't add them all on—so what is so special about this one, beyond just that it tested well, and moreover what is so fundamental about it that it has to be integrated at the browser, rather than extension, level?

This might also be good for Pocket, but I don't think that had anything to do with Firefox's decision. Why would it?

I think that I was unclear. When I said "best for widespread adoption of Pocket", I meant just that: Mozilla have decided, after gathering data, that Pocket is a service that is good for users, and they want as many people as possible to use it. I did not mean to suggest that this was an attempt by Mozilla to promote Pocket for Pocket's sake.

Oh ok, sorry for misunderstanding you before.