Optimize our meta tags?

Question

First off I should say I'm ignoring the following meta tags: beginner, homework, reinventing-the-wheel, and most with less than 100 views.

---

A short overview of the discussions around these tags is outlined below:

1. Tag synonyms (optimize -> optimization)

   We synonymized optimize ->optimization.

2. status-completed - Tags for [optimization], [performance], [speed]

   This came about as there was a synonym request between performance and optimization. We merged speed and performance. However it was raised that optimization should instead be three tags; speed-optimization, memory-optimization and code-size-optimization. Only the last tag seemed to have disagreement.

3. status-planned - Should we just put an end to [optimization]?

   This suggested removing the optimization altogether, in favor of other tags. Such as algorithm and performance.

   Again splitting this tag into three x-optimization tags was suggested, heavily upvoted and seems to be the plan.

4. [tag:optimization] was made a synonym of [tag:performance]

5. Optimizing [performance]

   This requested that optimization be merged with performance. But we don't think we should yet. It also seems like we know there are issues with this tag, but don't know how to deal with them.

6. Time limit exceeded and performance: what's the difference?

   We agree that these are not the same tag, and so shouldn't be synonymized or merged.

7. Would burninating [performance] bring good performance?

   Where we came to the conclusion that it's a useful tag. And points to three ways to optimize the code; performance, memory usage and readability.

And so we now have the tags: performance, optimization, time-limit-exceeded, complexity, memory-optimization and cyclomatic-complexity. I'm not sure if algorithm fits in this too.

---

And so I'd like to request we do the following:

1. Normalize the naming of the tags. This sets a clear naming scheme to follow when adding similar tags. And also helps users find the correct tag when searching for 'optimization'.

   • performance ->speed-optimization
     Additional synonyms performance-optimization
   • complexity ->complexity-optimization
   • cyclomatic-complexity ->cyclomatic-complexity-optimization
   • memory-optimization: no change
   • Should we rename time-limit-exceeded?
   • Should we rename algorithm?

2. Should we add more of these tags?

   • code-size-optimization
   • readability-optimization

3. How should we deal with optimization?

Answer 1

Should we rename algorithm?

I think we need to just junk it. It's not used very well (it's supposed to indicate an interest in the algorithm design, but it tends to be added by everyone implementing a well-known algorithm). I don't think the tag is adding any value to the questions that have it at present.

Answer 2

Yes, we should rename the tags

I'm ignoring time-limit-exceeded, algorithm, code-size-optimization, readability-optimization and optimization in this answer.

Pros

• Clear naming scheme if we add more of these tags.
• Means people that search for "optimization" can see all the different tags and pick the one that is most relevant.
• Adding tag synonyms that follow x-performance will increase the discoverability of these tags for existing users. Currently performance is our 6th biggest tag and so it's common that users will search for it when adding a tag.
• Performing a tag-rename and synonym is quick and easy.
• This should increase site quality just from appending "optimization" to the end of the word, as it'd mean the tag name is all the description you need. Take the latest complexity question, it has nothing to do with big O, and definitely doesn't have anything to do with optimization / improving performance.

Cons

• Renaming the tag may cause some questions to have incorrect tags.

  Since we'd be renaming the tags saying inline with what the tag already means, these questions would be incorrectly tagged already. We aren't changing the meaning of the tag.

  This means the rename alone doesn't change the quality of the site. But it set us up to more easily recognize these incorrect tags on old and new questions and so will have a net positive outcome.

• As pointed out by Toby Speight, having complexity-optimization and cyclomatic-complexity-optimization may be confusing.

  As pointed out above I think this is because the name "complexity" is vague, and so we could use big-o-complexity-optimization.

---

And so I suggest the following renames: (all tags fit in the 35 char limit)

• performance -> speed-optimization
  Additional synonyms: performance-optimization
• complexity -> big-o-complexity-optimization
  Additional synonyms: big-o-complexity-performance
• cyclomatic-complexity -> cyclomatic-complexity-optimization
  Additional synonyms: cyclomatic-complexity-performance
• memory-optimization
  Additional synonyms: memory-performance

Answer 3

I think that complexity-optimization might benefit from a rename (e.g. computational-complexity-optimization) to distinguish it more obviously from cyclomatic-complexity-optimization. That seems pretty wordy, but I can't see a good way to solve that.

Answer 4

We can do away with most of the tags which are up for debate, without problem. What are those tags doing there anyway? Problem description should be in the question itself anyway and meta tags are usually quite fuzzy.

Tags can be used to search on, to select and/or ignore. So it makes somewhat sense to have beginner, homework, reinventing-the-wheel, programming-challenge and such. There are people who are above averagely interested in them who can follow such tags, and people who are absolutely not interested in such questions but see them as clutter can ignore them. Perfectly fine.

But who's going to search for performance or complexity? I don't see a use-case for them. Are experts going to be interested in such tags, like they could with beginner or homework? Unlikely. Are the tags misused a lot? Absolutely.

memory-optimization should stay around since it appears to be a bit of a niche in modern programming (with the often related embedded tag), but the rest, I wouldn't miss them. Do away with this discussion once and for all and nuke the tags.

Everybody wants performance. Making it explicit in a tag doesn't help anything and should such a question contain a request for it, just having it in the question body would be enough, I think.

Answer 5

(I'm not sure whether this is right, so use your votes to indicate what you think!)

When we see time-limit-exceeded, it usually indicates a failure of an algorithm to scale well (assuming that the code is sufficiently tested on small inputs, so it's on-topic). So I'd rename those complexity-optimization as proposed. We'd need to keep the existing tag as a synonym, of course - this is a tag that brings many new members here.