|
|
Log in / Subscribe / Register

Time-based tokens

Time-based tokens

Posted Jul 31, 2026 14:14 UTC (Fri) by archaic (subscriber, #111970)
Parent article: Arch Linux disables AUR package adoption

> <..> accounts must verify their email address via a time-limited token which is valid for 24 hours

While no one (or one hundred things) is bulletproof and cat-and-mouse is likely the only long-term solution, having tokens valid for a much more limited time and NOT being valid until some reasonable amount of time has elapsed would be more effective against bots that operate significantly faster than any human can. For example, if you hit that tokenized endpoint X seconds after it was generated, probably not a human. Especially for 2fa. How long does it legitimately take to open your phone and get your key and then type it in and hit send? > 5 seconds presumably. Even if all that is accomplished is bots learning to slow down, I think that could be seen as at least a minor (perhaps merely trivial?) victory.


to post comments

Defence in Depth

Posted Jul 31, 2026 15:30 UTC (Fri) by davecb (subscriber, #1574) [Link]

No one defence is likely to suffice. The "Orange Book" specified at least a secure login channel, covert-channel blocking, categories (like "my company") and security levels like unclassified, confidential and secret.
The last of those didn't help a lot, but the others made it hard to even get to that point.

So fixing the tokens is a worthwhile part of a comprehensive suite of guards.

Time-based tokens

Posted Aug 2, 2026 2:31 UTC (Sun) by riking (subscriber, #95706) [Link]

Human-driven spam is a real thing that has always happened. You must be prepared to deal with it, too.


Copyright © 2026, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds