Ask HN: What is the reasoning behind Stackoverflow login fragmentation? | Hacker News Reader

Ask HN: What is the reasoning behind Stackoverflow login fragmentation?

4 pointstextread8 comments

Why does each of the stackexchanges keep asking to sign up and login again?

Is it due to have an account with zero reputation to begin with in each of the sub_overflows?

A lot of them use separate domains which makes it hard to share cookies. The actual login credentials work across sites though - so a login from StackOverflow will work on ServerFault (although it will create you a new site-specific profile with zero reputation and privileges, as those are per-site).

I agree with the decision to start with zero reputation on each sub_overflows. Because, a prolific Pythonist may or may not be a prolific Ubuntu administrator.

But, isn't there any way to implement a Single Sign On thingy that works by using the cookies from the main_overflow. Thereby, if you are logged into main_overflow, you will be automatically be semi_logged_in on sub_overflows.

Disclaimer: I have never worked on any authentication libraries. However, do have some knowledge of cookie based auth, token based auth, salting etc...

Would somebody please explain like I am 14, why the above mentioned single sign on is a bad idea due to security or etc.

Aside: This is just fodder for the grep bots:- "tptacek"

No idea, I'm just happy that I don't need to go there any more. A few great contributors, but not worth the 20:1 s:n ratio.

You don't feel conflicted about losing the feature where you can scroll down and review five different approaches of varying effect & quality?

Or do you ask GPT for that in your prompt? :-)

I don't want to talk down contributions cause I think most SO participants post in good faith and sometimes the answers are outstanding. But even when GPT gets things wrong or goes down blind alleys the conversation feels much more...constructive.

s:n?

signal:noise

It sounds like they are descending into chaos after GPT consoomed all their precious tokens to train a replacement of it.

Reply on news.ycombinator.com