We had some great sucess with eliminating the pointless tag file recently. I would like to continue the quest to rid the world of bad tags by proposing a new candidate for burnination: token. It is used for a multitude of purposes:
- CSRF tokens
- OAuth tokens
- Authentication tokens
- Session IDs
- The beloved South Park character.
Just like file, the concept is both broad and vague. No one would say "I am a token expert" on a job interview. No one would say "Hey, I have a question about tokens" to a collegue. The natural follow up question to both is "Well, what kind of token are you talking about?".
There is a usage excerpt, but it was added recently and does not reflect how the tag is used:
An "authentication token" is either a hardware device, or piece of data used as part of an authentication mechanism.
So, this is what I propose: Questions about CSRF tokens should be tagged with csrf. Questions about OAuth tokens should be tagged with oauth. I think you see the pattern here. Questions about X-tokens should be tagged with X, not token. (Actually, I think a lot of these questions are tagged with token just because the OP wanted to tag with "X token" but didn't know that you can't have a space in a tag name.) Questions about generating secure random tokens should be tagged with random. If there is a need for a tags matching the current tag wiki quoted above, they should be called authentication-token and/or hardware-token.
We are dealing with a lot of questions here - 110, and counting. If we do this, we should do it slowly so as not to spam the main page. It does not matter if this takes some time. It is also encouraged to fix other flaws in questions while editing them.
So, what do you think? Should the tag be burninated? If not, what should be done to fix it? Any additions or changes to my retagging suggestions?