First, we have to decide what we mean by typo questions. Some have suggested that syntax error questions count as typo questions, while others believe it's really for questions where the OP forgot how to spell their own name.
Syntax errors as "typo questions"
Syntax error questions historically happen because the user did something wrong. This is often followed by a sometimes cryptic message from the compiler (seriously, SQL Server).
If you're really raring to close syntax error questions as too localized, good luck. There are only about 41,000 of them. I do not believe the ones that exhibit certain characteristics should be closed as too localized -- if they are duplicates, close them as such, but not as too localized.
Too localized has commonly been used to address questions that just aren't helpful to future visitors. On their face, you may think that a syntax error question falls right into this bucket. After all, Who else will ever have a table named SAP, anyway?
But that's not why these questions are useful. Arguably, the syntax itself isn't, it's the conditions that cause that problem that are useful to future visitors. In the above example, the user's question exhibited those characteristics that make it a passable question:
- They included the code that caused the problem
- They included the error message in all its glory (good thing to search on)
- The title is actually better than most titles dealing with these sorts of problems in that doesn't say, "What's wrong with this code?"
In short, if it weren't for the fact that this is a syntax error, we could otherwise agree this question is much better than others we see (and better than others that stay open).
You may still believe this question doesn't provide value -- and you're right, by itself it does not -- but, it provides the foundation for value (it gives an answerer everything they need to knock everyone's socks off with a really good answer to an otherwise mediocre question), and that's important.
Take a look in the SQL Server question I linked to, there are the beginnings of really good answers, notably in the comment that says:
You might want to add some syntactic sugar and specitfy the joins as
INNER JOINinstead of just
Join. Won't make any difference, but it's arguably easier to read. -- jpw 10 hours ago
Besides that comment (which should really be fully fleshed out into an answer along with fixing the typo) you also have the ability to take what was provided and turn it into something teachable. For instance, what does the error message:
"Msg 4145, Level 15, State 1, Line 6
An expression of non-boolean type specified in a context where a condition is expected"
mean? There are several parts to it that we normally forget about, but "msg 4145, Level 15, State 1; those are all things that if there were an answer around, would teach future users something about SQL Server they probably didn't know -- and we haven't even gotten to the error message itself! If the answer also included what "an expression of non-boolean type specified in a context..." meant with examples, it would help future users understand something that is hard for a beginner to understand.
That's the value here, and it's really just sitting there, waiting to be answered.
Once we have that good answer that fully fleshes out these things, then you can start searching for other questions with that same error message, and see if the answer helps the other questions (undoubtedly it will help most of them, if it's fleshed out like I contend it should be).
Contrast that with what the claim for the desired endgoal is:
Close and delete all typo questions.
Is the contention that there should never be any IndentationBlock error questions on Stack Overflow? After all, that only happens in Python when the user does something stupid (the python equivalent of a syntax error). Tell me: If we close (and subsequently delete, or even just close all new ones) how will we ever give someone the opportunity to post that answer that really settles the issue? Then, users will be back to searching google for that error message, and coming up against (Uggh) Forums.
That's question I talked about above is the type of question I want to save. There are specific criteria for it being salvageable. If it's a duplicate, close it immediately.
I forgot my own name - typo question
When a user comes back and says "Hey, I forgot my name was Rob, and wrote Bob", in their question, and there's no answer that has upvotes, I'll close and delete the question. If there's a good answer (that teaches something outside of "your name is Rob, silly") then I won't delete the question.
If you're willing to concede there are typo questions that are salvageable, I'm willing to concede there are some that are not.
What to do about Questions that would have been closed as 'too localized'
Even though I'm against custom close reasons, I do believe some should be closed. They should not be closed as "typo question", because that just goes down the same hole we just dug ourselves out of. Rather, they should be closed as
"Questions must demonstrate a minimal understanding of the problem being solved."
This is an accurate close reason for what is coming across. I argue that if you misspelled your own name, you don't have a minimal understanding of the problem being solved.
Now that "Minimal Understanding" has been eradicated due to abuse; I would vote to close these questions as "Questions asking for debugging help must contain a useful problem statement." when there little chance the question as written will help anyone else.
All too often, these questions do not include information that a person searching the internet would search for to get help for their own problem.