Although most people on HN and the like will usually opt out or refuse such a request, many people will not care about the privacy/security issues if it means less work for them.
So people get burned by some Nigerian scammer, decide never to do any business online again, and what good does that do for the YC crowd?
Second, there are lots of convenient features that we don't build because they send a terrible message. For instance, it would be really convenient to work with your bank via email.
Third, he's right: many of these services probably do bank your password, which is an absolute nightmare. Almost every web app is going to lose its database to SQL injection at some point in its lifespan.
And the contacts API that the major providers are giving are recent. It's not like services like Yelp and FB had the choice before.
Ps.: Jeff Atwood's "from the hip" writing style shows. I'm yet to find one single post from him that seems slightly insightful. Instead of spending so much time ranting, he could spend a tenth of that time thinking about why Yelp keeps that "feature".
Edit: could someone please explain me why of such aggresive downmodding?
You said that Atwood should have spent time thinking instead of just ranting, but then you kind of did the same thing....
And that's not to mention giving the "they're a company, anything goes" and "it was the only way back then" excuses; those are meaningless.
The same goes for websites that do such thing. They know that some users might be annoyed by such "feature", but the amount of new users that will be exposed is much higher, so they keep doing it.
What I find hard to believe is that the ones that complain about these kinds of mechanisms don't understand that themselves are not part of the target of said mechanism. Instead of simply ignoring it, they feel "victims" of it.
And I'd like to know where I said "anything goes". If that was the case, I would say that scammers are right to do what they do because they never forced the users to give sensitive data. What I did say is that, if a webservice wanted to provide a way for the user to (semi-)automatically invite people in his contact list, the only way to do that required email and the password.
Better now?
What I find hard to believe is that the ones that complain about these kinds of mechanisms don't understand that themselves are not part of the target of said mechanism.
These mechanisms are prominently shown to all users of the web site who want to import contacts, whether they are technical or not. There isn't a banner above them which says, "this feature is aimed only at people who don't understand the security implications."
Making money by teaching someone to compromise their security is a morally bad thing to do, whether that person knows they are being taught to put themselves at risk or not.
There isn't a banner above them which says, "this feature is aimed only at people who don't understand the security implications."
If you do understand the security implications, then the banner is irrelevant!
Also, they are not saying that you must use the feature either, are they? What happened to freewill? Does everyone think that you are supposed to consume every feature of the service just because it's free and it's there?
And to say they are "teaching to compromise security" is a gross dramatization. They are not teaching anything at all.
I agree that some time ago, it may have been a (dubious-yet-) reasonable feature. Now it's just irresponsible.
... until it bites them in the ass.