When adding tags to a question, there's a neat box listing all the potential tags. The first tag in this box is the closest match, which is the shortest tag with the currently-typed string.

Often, this tag is a substring of the intended tag. Additionally, sometimes the second part of the intended tag is itself another tag. For example, when adding , a user will encounter first, and then . Just over 90% of the time the user gets it right and adds . 10% is way too high an error rate for this process.

Here's some example cases of this problem:

[Foo]    | Count | [Bar]     | Count | [FooBar] | [Foo][Bar] | [Foo][Bar][FooBar]
asp.net  | 123k  | mvc       | 14.4k | 34.7k    | 3,283      | 557
batch    | 3.2k  | file      | 13.4k | 2.7k     | 153        | 15
embedded | 2k    | linux     | 31k   | 337      | 241        | 22
linux    | 31k   | kernel    | 1.4k  | 2.1k     | 575        | 93
session  | 9k    | variables | 4.5k  | 708      | 141        | 3

...and that's enough random data-finding for me. This list/chart would be fairly easy to expand with an automated tool.

I humbly offer that most every use of [Foo] + [Bar], when [FooBar] or [Foo-Bar] exists, ought to be replaced with (or at least evaluated for replacement with) [FooBar]. I less humbly suggest that every use of [Foo] + [Bar] + [FooBar] ought to be reduced to [FooBar].

  • Does [foo] [bar] always equal [foo-bar]? Or are there a significant number of examples for which the separate [foo] [bar] tags actually have a different and distinct meaning from [foo-bar]?
    – user102937
    May 16, 2012 at 22:28
  • 2
    @RobertHarvey - I'm unware of any such examples. I compiled this list by manually looking through some tags; it's a pattern that I've seen before. I'm not good enough with SQL to do this programmatically in Data Explorer. If you or someone else could generate a query, I'd be happy to do my part in reviewing the list. May 16, 2012 at 22:38


You must log in to answer this question.

Browse other questions tagged .