126

I appreciate the benefits that the quality filter brings to the site, but I've noticed that posts are still making it through with a lot of lower case 'i' instead of 'I' and 'u' in place of "you".

I understand the acceptable practice of using abbreviations while texting, and I accept that these uses are permissible when scratching out a quick message to a friend or loved one that you'll be late for the movies, but I think the 'i's and 'u's detract from the professionalism of the questions and answers on the site. I think there are many that agree with me in general terms, as we wouldn't have the filters in the first place if this weren't the case.

I don't know how the filters work (I can make some educated guesses, obviously), but I don't need to know. I'm asking if it's possible to strengthen them just a bit more against these two cases. I don't mind editing posts where I see 'i' and 'u' being used, but I think helping people fix their own posts at the point of origin is even more powerful.

As has been brought up multiple times, "X Language and Usage", as well as upcoming "Stack Overflow (in X language)", sites are vital to the network, and some of these languages contain constructs using 'i' and 'u'. Obviously, a filter containing such a rule on these sites would be counterproductive, so I'm proposing this for English language sites only.

Improve this question
15
  • 8
    I've thought this myself. A solitary i with no surrounding formatting and not followed by a . or ) will mean I what, 9999/10000 times?
    – user154510
    Sep 6, 2011 at 2:06
  • @Matthew Yes, I agree. I was under the impression that the heuristics skipped the code blocks, which should be the primary place where you run into 'i' and 'u' by themselves.
    – jonsca
    Sep 6, 2011 at 2:08
  • 10
    @Matthew Unfortunately that 1/10000 times will probably be in badly formatted code samples and fixing those will be a PITA.
    – Adam Lear StaffMod
    Sep 6, 2011 at 2:09
  • @Anna I think there's already a mechanism to pick up poorly formatted code (?), so I suppose it could work in tandem with this proposed improvement.
    – jonsca
    Sep 6, 2011 at 2:11
  • 8
    @Anna: Does poorly-formatted code really become that much worse if a variable's name is changed from i to I? It won't change the meaning of the code, and it has to be cleaned up anyways.
    – jscs
    Sep 6, 2011 at 2:25
  • 2
    There is a (somewhat) legitimate translation misunderstanding that can occur because the first person singular pronoun is always capitalized in English but in other languages follows the same capitalization rules as other words. That doesn't make me any less in favor of this idea, and writing "u" for "you" is still basically inexcusable.
    – jscs
    Sep 6, 2011 at 2:30
  • When I type a lowercase i it is because I spend a lot of time in the VS editor, Word or Outlook, all of which correct it for me. Despite it being technically incorrect, I believe most people can read it just fine. I'm with you on the u though - abbreviating words is just lazy. (Notice how I managed to uppercase my I's for the sake of this comment).
    – slugster
    Sep 6, 2011 at 2:35
  • 3
    @slugster I would say the key to your notion is that it is being corrected. You are seeing after the process even if you aren't typing it. The end result is that your colleagues get emails from you that look more professional than they might had the software not fixed it. Since there's no autocorrect here, I would think you would want to still see through the "duty"(loosely) of capitalization.
    – jonsca
    Sep 6, 2011 at 2:42
  • 1
    Now you've just made me mad that search is blind to capitalization, so there's no easy way to find posts that contain "i" and "u."
    – Pops
    Sep 6, 2011 at 3:00
  • for(int i = 0; ... Surely there are more cases that would have to be considered?
    – jball
    Sep 6, 2011 at 5:17
  • 1
    @jball If it's in a code block, I don't think it gets considered anyway. The only concern would be if someone said something like, "i was varied from 1 to 10" within the text, as Kit has pointed out in the answer.
    – jonsca
    Sep 6, 2011 at 5:36
  • 2
    thinkgeek.com/tshirts-apparel/unisex/generic/8298
    – user541686
    Sep 18, 2011 at 8:06
  • @jonsca: would a spelling filter work better in this sense? Some sort of parser that checks on the spelling quality standard? One could then review/reject the proposed amendments
    – ElCid
    Aug 23, 2012 at 10:11
  • @ElCid I think that would be an awful lot of overhead (having to load dictionaries, etc.). This would just be a simple heuristic and require no intervention.
    – jonsca
    Aug 23, 2012 at 10:15
  • Could we just get one of those fonts that make every letter uppercase? Just kidding: we would use 90% of our users. :D Related xkcd
    – Anonymous Penguin
    Mar 20, 2014 at 16:11

5 Answers 5

Reset to default
8

I may be a bit biased here (if you examine my posts, particularly those from my first year of activity on SO / MSO, you'll find a lot of lower-case is), but putting undue weight on this particular aspect of a post seems misguided. There are a lot of mistakes that are both more distracting and harder to correct.

The use of "txtspk" - including i and u outside of code blocks - does factor into a post's quality score; if it's especially frequent, such language can block a post entirely, or send it into the Low Quality review. In practice, it'll likely be combined with other indicators that'll either redeem or damn the post.

I see no reason to reject a post purely for the existence of these characters, particularly given the likelihood of false positives.

Improve this answer
1
  • 1
    I can forgive your youthful indiscretions. Obviously, you know the Colonel's secret herbs and spices that go into the quality filter, but it seems as though it is doing a better job with it now, anyway. I think there are some users who write eloquently and use the 'i' for stylistic reasons or because it's not a cardinal sin in their country of origin, and that's fine. I think my underlying theme was that people should take pride in their own posts, rather than relying on editors who only have a finite amount of time and energy, though perhaps at first blush it didn't come across that way.
    – jonsca
    Feb 21, 2013 at 0:21
36

(╯°□°)╯︵ ┻━┻

Improve this answer
3
  • 10
    The cloud creature fell off of his single ice skate? I was kidding, btw :)
    – jonsca
    Sep 7, 2011 at 23:24
  • 14
    Hey, if you're not man enough to admit you want to flip over a table when you see bad grammar, I understand. It's cool, it's cool.
    – Pollyanna
    Sep 7, 2011 at 23:32
  • 1
    I would want to flip over a table. Still, I like my interpretation better.
    – jonsca
    Sep 7, 2011 at 23:35
19

i in an English sentence should be I, but in SO at least, it might be the author's intent to reference a variable, and said author might not be back-ticking (which I can't demo from my iPad's keyboard, because it doesn't seem to have a ' (apostrophe) pointing in the "wrong" direction) appropriately.

Improve this answer
2
  • 30
    Hold your finger on the apostrophe, and you'll see the backtick key appear.
    – Kevin Yap
    Sep 6, 2011 at 4:22
  • A valid concern, but probably one that a heuristic can be built against (placement in the sentence, proximity to other words). The actual filter algorithm is not public, so I have no idea what the mechanisms are now, but there may be room for adjustments.
    – jonsca
    Sep 6, 2011 at 5:34
17

I'm in two minds about this. The thing is, in my experience lower-case i, and use of u are excellent indicators of overall question quality. Do we want to automatically remove the things which allow us to detect crud?

What I mean is, that in general correcting these 'errors' is not going to bring more quality questions to SO.

This may seem elitist, but I don't think it is. Free Q&A is about each party doing their 50%, and for me, someone who can't even be bothered to use a shift key or type two extra letters is someone who is most likely not going to take the time to formulate their question decently, or even think about it much at all.

So yes, u & i can part ways, they can just go away altogether :)

Improve this answer
2
  • 3
    I like the idea of using this to detect crud. If this feature were part of the quality filter, I assume it could be incorporated as a criterion into the post quality score and land such posts on the "Low Quality" page in Review. Human editors would lose this edge, but it's regained in using an existing automated tool.
    – jonsca
    Oct 19, 2011 at 10:58
  • I find myself incapable of imagining a situation where I would agree more. If someone cannot be bothered to take some time to formulate a decent question, why should I bother to help them? +1
    – bitmask
    Sep 5, 2012 at 0:30
12
+100

With the emergence of non-English sites (currently only German Language & Usage and French Language & Usage are relevant), a filter of this kind would have to be tuned to the site's language(s). There are languages where i or u is a word (not German or French, but some of the languages with a site proposal about the language itself or some other topic).

Improve this answer
7
  • Yes, certainly I hadn't specified as such, but of course the filter would have to take locale into consideration. Some of the other commenters have raised a valid point, grammatical principles in other languages can dictate very different capitalizations. I'm mainly drawing on experiences on SO and SU.
    – jonsca
    Sep 7, 2011 at 21:54
  • 1
    I think they already are; a lot of filters seem to be a lot stricter on Stack Overflow (they might just be disabled on other sites).
    – Aarobot
    Sep 7, 2011 at 23:29
  • @Aarobot I know some rules are applied even on sites where they're inappropriate, so I tend to be wary.
    – Gilles 'SO- stop being evil'
    Sep 7, 2011 at 23:37
  • If you add a bit more to your answer about the proposal at hand, assuming it's limited to English sites, that would be excellent.
    – jonsca
    Sep 24, 2011 at 10:40
  • The only language where either is a word (to my knowledge) is Italian (currently 39% committed), where 'i' means 'the'.
    – Kevin Vermeer
    Oct 19, 2011 at 11:17
  • @KevinVermeer There are plenty more, including several languages where both u and i are common words (Catalan, Czech, Polish, …).
    – Gilles 'SO- stop being evil'
    Oct 20, 2011 at 2:30
  • @KevinVermeer Just as English switches from a banana to an apple as a sandhi effect, Spanish for the same reason switches from o to u for “or” when the next word begins with a phonemic /o/ monophthong. So for example it’s cero o uno but uno u otro, and hoy o mañana but mañana u hoy.
    – tchrist
    Jul 19, 2015 at 1:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .