As part of our closing overhaul, we've moved, reorganized, and renamed a number of close reasons, including too localized. While that change was informed both by widespread misuse of the reason and numerous meta posts from you, many of you have also indicated concerns about its new home in the off-topic menu.
And when you're concerned, we usually are, too.
So, we wanted to provide a little more background, and more importantly, set up a place where you can share any gaps you're finding once the changes roll out network wide.
- If you find examples that TL used to cover, but that can't be addressed by the new "off-topic" reasons, post them as answers here.
- If you want more background on why something needed to be changed, read the large number of words below.
Did TL clear out many bad questions?
Sort of. It definitely helped us eliminate some bad questions, and we need to make sure we deal with those, but it was the least frequently used close reason - a recent sample found it used on roughly 1.3% of total questions asked on Stack Overflow (with similar network stats). But reviews of sampled questions by mods and staff found that roughly half of TL closures probably should not have been closed, meaning:
- its correct use was affecting roughly 0.65% of questions (for perspective, all closed questions were about 12% of those asked)
- in its current form, it seemed to be causing as much harm as good
- on average, you'd have to read just over 150 questions before you encountered a single one of these.
That's not to say that we should ignore those questions or don't care about closing them, just that the total volume they represent, while not trivial, also... ain't overwhelming. But the real issue is that it wasn't working consistently:
What was wrong?
The two main problems with the old reason were:
- no one really agreed on what it meant, and
- it attracted false positives from trigger-words (locations, etc.).
Its description had three parts, and the first two parts were very broken:
only relevant to a small geographic area - there are extremely few questions that need to be closed due to locational constraints that aren't closable for other reasons, and no one ever agreed on what "small" meant, or even what units to use. This was a distraction from the real need.
limited to a specific moment in time - again, there was no even semi-consistent standard. Some people felt that meant days, others months, etc. Some sites applied it to any beta, which seems reasonable, until you remember that gmail was in beta for five years. Questions that are obsolete definitely need to be addressed, but most technical questions will eventually suffer that fate, so making everyone pick their own definition of how short is too short before that happens wasn't working.
The third part was more generalized:
- unlikely to help any future visitors… or an extraordinarily narrow situation not generally applicable to the worldwide audience of the internet This is starting to get to the heart of why TL was important. Now, the old wording did invite debate over when to use it - the first clause seems to require just that one person other than you could benefit (2 people total), while the second one suggests that some meaningful percentage of all internet users should. (As of today, 1% of all internet users would be 24MM people).
But setting aside the lack of a clear line, isn't that concept useful? It is. The core of it sounds like the generalized "too narrow," which some have suggested should be a network-wide reason like "too broad".
But "too broad" has a line we're comfortable with: "Answerable in several paragraphs." There's still some room for interpretation, but we're all on the same planet.
But, while we do really like the idea of "too narrow," because it speaks to our belief that the best questions are those that benefit more people:
"Too narrow," is very, very hard to establish even a rough guideline for.
Which is why our top users are still interpreting it wildly differently, even within individual sites.
How do we retain all the good stuff
Too Localized was doing something important. It was being used, quite helpfully, to allow sites to cull the specific types of one-off questions that their experts didn't want. The most common of these, by far, was
- "code dump"/"find my typo" questions on Stack Overflow - where the author provides a huge block of code, or a broken site's url, with little more than, "Why not working?"
They should be closed. But it's actually a lot easier to close them as a specific off-topic reason, because it eliminates all those silly debates about whether someone else might make the same typo, or how many helpees are required. Instead, we say, "our definition of our topic has explicitly excluded that". We're not debating anything, our store simply doesn't sell that thing:
We've refined our definition of what's on-topic here. The "headline" may be 'programming', but what we mean is:
- Programming, but not Code dumps.
- Programming, but not whiteboard problems.
- Programming tools, yes.
- Programming snacks, no.
It's really not all that different than the way other sites, over time, identified the specific question types they don't want, and made them off-topic:
- Gaming, but not "name that game"
- photography, but not "fix this picture"
- cooking recipe replacements, yes
- cooking recipe requests, no
Most of those could fall under Too Localized, but it actually makes more sense to make them part of the definition of the topic, because it makes their closure less subject to interpretation.
What now? Post any gaps as answers here.
We hope the above helps make some of you less worried, as we do think this will actually work better for the key good uses of TL.
But we're still a little worried about gaps.
If you find a question that previously fell under TL, but for some reason can't easily be dealt with through off-topic, post it as an answer here, so we can follow up and adjust as needed.
For sites other than SO, you may want to discuss it on your local meta first, to get community consensus, but that's up to you.