Is there any way that we could somehow decrease the sensitivity of the spell-checker? There are so many false positives. You can be patrolling a batch of 40 and only 1 or 2 will actually be misspellings. The others will be something like the name of an app, or a medication, or even something in a foreign language but just aren’t misspellings. Any thoughts?

If you click “No” on the suggestion, for all I know the tool “learns” the word isn’t a misspelling and won’t suggest it’s a misspelling again. Although I do see why it’d be frustrating to have to constantly click that “No” button because it keeps bringing up not-so-misspelled misspellings. Ah well. I bet in future years it’ll have paid off.

It’s been my experience that clicking “no” doesn’t help much. The same correctly spelled word will pop up again many times in some cases.

I agree. I’ve been chased away from spell check many times, and I am afraid to go back… We should really do something about it.

Chased away from the spellchecker… ? Sounds like we’ve got a real problem in the tool then!

Yes. We should be changing the spellchecker. I have also been chased away from it.

I wish there were some kind of magical sensitivity dial for the Spellchecker - that would be pretty neat:slight_smile:

But I’m afraid we’re not holding out on any secret engineering tricks there! It’s just that the Spellchecker can only tell what might be “wrong” by comparing the given words to a set dictionary of common words/terms, and the voted-on whitelist. So things like brands, foreign words, unusual scientific terms, etc - they’re just not going to be recognized as correct right off the bat. It takes several votes of “No” before they’re put onto the whitelist, so even if you vote No, you might see that vote recur until it gets enough votes to go onto the list; that’s expected, but I can understand it feeling repetitive at times! 

Some Spellchecker users are spot-on every time (cough @Donagan cough), and if everyone were like that, we might be able to whitelist words after a single vote… but other folks (particularly newer users) do end up “fixing” or voting in ways that are seemingly pretty random, so it’s tough to find that right balance when you have all sorts of people using the tool. Any boosters and admins who used to approve/reject words on the whitelist manually (back when it worked that way) can attest to the fact that one vote didn’t always get things identified correctly for whitelisting. Do you guys have any ideas about how to get people not to approve incorrect words? I can’t promise any big engineering changes on the tool in the near future, but I can always save any good ideas in a brainstorming list for if/when there’s ever another big rethinking or recoding of how the tool works, in the future. 

I don’t totally understand what you mean, @Glitterunicorn and @Minecraft.34 , by saying that you’ve been chased away and are “afraid” to go back to it, though? Do you just mean it was boring for you, or did something actually make you afraid? Of course, if you don’t find it fun, there’s no obligation to use it. That style of vote tool isn’t necessarily up everyone’s alley - perhaps you’d prefer the Copyedit Greenhouse if you want to do more self-driven spelling corrections? That can be a better one if you don’t like voting all the time and just want to focus on fixing.

Just for a retrospective giggle, it is worth noting how much the Spellchecker actually has improved over the years… Believe it or not (for those of you who are newer to the tool!), there was a time when the Spellchecker highlighted all URLs as potential errors, and when it was tripped up by any special characters (so, for example, it highlighted every word with an accent or similar unusual character). In that sense, the sensitivity has been adjusted down, in that the tool was rewritten to ignore those things. It’s better than it used to be, even if it’s not ideally all-knowing. But then, if it was all-knowing… well, we wouldn’t really need the Spellchecker tool! If the software could identify all mistakes correctly every time, we could likely just make the fixes, programmatically, with a bot. The beauty of the tool is the idea of combining some automatic detection of potential errors with the the spelling gurus and wiki-brains in the community to get the right end result:slight_smile:

@Anna you must’ve made a mistake. I never commented on this thread. But it’s okay.:slight_smile:

Oh sorry about that @Gymnast-7 - I had multiple threads open and copied the wrong name in tagging. I’ll fix it:slight_smile:

It’s okay. The other persons username sort of sounds like mine anyways. : P