Good Tags now save a lot of work later!

Question

Since you are just setting out with this SE, you might appreciate to hear about my learning experience: I've been a moderator on Bitcoin.SE for more than two years now, and I have spent a lot of time on fixing tags there: Tagging can quickly grow rampantly, when not cultivated.
E.g. when I started the clean-up, we had lots of

• meta-tags (describing the type of question instead of the topic), e.g. definition, opinion, comparison
• ambiguous tags, e.g. future, computing, offline
• too broad tags, e.g. mining and transaction were on more than 10% of our questions
• duplicate tags, e.g. jargon, terms, vocabulary, and terminology

---

So, perhaps consider these ideas to evaluate tags:

What properties do good tags have?

1. Descriptive. Does the tag add valuable information when it is added to a question?
2. Representative. Can the tag be the only tag of a question?
3. Unambiguous. Does the tag commonly mean the same thing for different people?
4. Confined. Does the tag create a specific, well-defined category?
5. Meaningful. Can you see anyone at some point subscribing to the tag or searching for related questions by using the tag?
6. Constrictive. Can you see anyone using this tag to ignore a topic?
7. Unique. Is there a very similar tag that is more popular which could be applied instead?

If a tag's usefulness is contested, think about whether its name or definition could be improved, but finally rather err on allowing it.

---

The good news is that your tagging looks pretty already. Lots of tags have descriptions and most seem to be good ones.

I'd just like to point out a few tags that we've already had run-ins with on Bitcoin.SE:

• mining and transaction ended up on almost 10% of our questions in the end. I suggest that you try to start out with a more diverse array of tags in those areas from the start. For some ideas see here: mining and transactions
• development tended to be misused for any and all questions about software development. Perhaps you would like to clarify it to monero-development or similar.
• We have a bunch of tags along the lines of merchants which we haven't found a good clean-up for, yet. Perhaps you can get a head start by offering a few more specifically defined tags such as face-to-face-payments, webshop-integration, and merchant-address-management.
• We found security and attack a bit too broad in the long-term and have since added more tags for specific topics such as majority-attack, race-attack, ….

There are few meta-tags such as usage, and technical-comparison.

And some tags may be too unspecific in the long run such as size, maintenance, integration, improvement, creation, support.

Believe me, it's much easier to do a little work now than cleaning up tags with 1400 questions later because the tag encompasses a huge part of all topics on your SE. ;)

I'd be happy to help or advise, if you are interested. I'm usually hanging out in chat, and will be looking at this SE from time to time. :)

Answer 1

Thanks again for the well research, Murch. I would like to comment on the four categories which you highlighted.

Ambiguous tags!

• ambiguous tags, e.g. future, computing, offline

I agree. This is the worst type and they should be removed constantly. If they just keep coming back (most of them wont), it should be considered to blacklist them.

Meta tags!

• meta-tags (describing the type of question instead of the topic), e.g. definition, opinion, comparison

I agree. There is a nice StackOverflow blog post by Jeff Atwood explaining the details and criteria.

Duplicate tags.

• duplicate tags, e.g. jargon, terms, vocabulary, and terminology

I agree. They are easy to identify and in most cases creating synonyms is sufficient.

Too broad tags?

• too broad tags, e.g. mining and transaction were on more than 10% of our questions

I partially disagree. And basicly just came here to say that, because I saw this thread. Quoting myself from the bottom of my answer over there regarding the two example tags:

And while we are at it: Don't be scared about mining or transaction at this point. In my eyes they are valid. On Ethereum stack exchange there are only 266 transaction questions (9.7%) and 194 mining (7.1%) questions. At this stage it might be also worth to read up through this meta thread from short after the Ethereum public beta launched, and also this thread with a good comment by Robert Cartaino:

Tags aren't like questions that have too many answers. They are supposed to organize somewhat large swaths of subject matter so folks can find and follow their general interests. Some of those interests are more specific while others remain somewhat general.

Answer 2

That's an interesting post. I've recently been annoyed at what I see as being useless tags: tags that are overlong, and overly precise. An extreme example being "human-powered-mining" (honestly, who's going to have the idea of searching for that ?), but also more subjective examples like "quantum-planning".

I see good tags as short (ie, what someone wanting to find questions related to a given topic would try out first) and in a kind of sweet spot being not too broad so it matches everything and becomes useless, and not too specific that nobody's going to search for it.

I think it depends on how people use tags. Do they type some keywords to search, or do they do find a tag list and click ? In the former case, short and non obscure tags are best, but in the second case, the long and obscure might do too (though if those are used, then the more generic ones should also apply, so someone looking for "quantum" also finds posts tagged "quantum-planning" as well as the dozen quantum-something that are bound to pop up.

I think what bothers me is tags that will be attached to a couple questions. I see those are pointless, unless the subject really is a specific one. But a dozen of "quantum-something", each matching a couple questions... I just don't see the point, it seems like tag spam to me.

Answer 3

At this point the generic transaction and mining tags are gone. I share the concern of user36303 that some of the resulting sub tags like human-powered-mining may never be used. We can always merge back the the overly specific mining tags later into a general mining tag if we change our mind. We cannot easily do the same thing in reverse (converting large numbers of general tags into specific tags). The downside of this approach is that edge cases (that do not fit any existing specific tags) like the one you mentioned human-powered-mining will keep appearing if we are eliminating the general mining tag as a catch all.

I am not sure which method (using broader mining and transaction tags like ETH SE tags or more specific tags like BTC SE) will work better for us long term. However the BTC SE method is clearly more flexible because we can merge specific tags into more general tags later much easier than the other way around.

Would testing both approaches be reasonable?

1. Stick with specific the tag replacements for mining and transactions (but trying to avoid creating more than we already have)
2. For other topics (not mining and transactions) we use general tags such as quantum-computing and not have any other more specific quantum tags.
3. Reevaluate in a few months to see if we prefer the results of 1 or 2?