Hacker News new | past | comments | ask | show | jobs | submit login

> Transparent updates, especially when they're nigh-totally transparent, are in an entirely different world than "do you want to update yes/no/cancel"

Transparent updates are great, until something goes wrong. What happens if someone manages to hijack the DNS entry to make 50 million Chrome installations download an update from a poisoned well? Or Google's quality control misses something and pushes out a buggy update that breaks Chrome's ability to reconnect and obtain a fix for that update? (This has already happened to companies the size of Skype and McAfee.) Or an update simply introduces a behavior that users don't want, like Microsoft using Windows Update to inflict WGA everywhere.

The net benefit of transparent updates is still probably greater than explicit updates, but both ways have costs and we shouldn't pretend that either is zero.




Don't those same problems also exist for explicit updates?

I suppose it's a mitigating factor on the part of requiring user opt-in, should a problem occur, but that's (IMO) only a fortunate side-effect. It's like saying that walking is better than driving, because if you run into something you'll only bruise your shins.


And opt-in updates also mean the ones that did get the faulty update hold onto it for much longer. ie, IE.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: