Recently, one of the "lifers" of History SE has engaged in mass removal of tags from questions.
Perhaps many consume content on Stack Exchange by visiting the front page of a particular SE site and glancing at it. However, some — myself included — consume Stack Exchange content by subscribing to tags. Without detailed tagging, subscribers may be left in the dark.
Since questions can have 5 tags, what exactly does minimalism in tagging hope to achieve?
Questions should have as many tags as possible and we should all continue to define new tags as needed.
A question having more tags is more searchable. Tagging is free; the limit of five encourages us to focus on the most significant ones. Tags are not mirrors onto the textual content of questions and cannot be equated to the appearance of a certain term in free text. Rather, they are editorial statements made in a structured form. As all our questions pertain to various named entities, and we want our tags to describe questions well, leaving tags off of them is a missed opportunity.
Tags that are more precise are richer in value because they are more distinctive. Happily, this site allows ranking users to create tags without prior discussion, which is the folksonomic approach also used on Wikipedia. Minor work on the taxonomy is undertaken by individual initiative, and major work by consensus. Editorial effort to change a question's tags or to combine two tags can support consistency and coherence; casting judgment on the usefulness of a tag's existence does not.
The inherent tension in information retrieval is a balance between precision (the fraction of documents returned which are relevant) and recall (the fraction of all the relevant documents which were returned). More precise search terms may have to be linked with a Boolean OR, while more vague search terms result in wading through pages of irrelevant results.
As a thought experiment, consider the difference in magnitude of the tag information between a question with two tags from of a vocabulary of ten terms, and a question with four tags from a vocabulary of ten thousand terms. Using a fixed-width encoding the former occupies 2*log2(10) bits and the latter, 4*log2(10000) bits; the second set of tags expresses eight times as much information as the first.
Let us make our tags as useful as possible by making them as informative as possible.
Our Help Centre contains the article What are tags, and how should I use them?. It defines tags as:
A tag is a word or phrase that describes the topic of the question. Tags are a means of connecting experts with questions they will be able to answer by sorting questions into specific, well-defined categories.
My view (with the usual caveat - please ignore any mod diamond you might see by my name), for what it's worth is that tags only add value to the site when used in this very limited way, or when they allow us to distinguish questions about a subject from those questions that simply mention that subject.
To (mis)quote Dr. Ian Malcolm: just because you can tag a question doesn't mean that you should.
Let me illustrate my point using the tags that were created (well before I joined the site) for the three main Allied leaders during the Second World War, fdr, stalin, and churchill.
Given that our two most popular tags (by a large margin) are united-states and world-war-two, it does seem likely that we will get quite a lot of questions about American Presidents, and that the President who led the United States through the build up to the Second World War, and through most of the war itself, might be expected to get more than most.
I'm guessing that was the reasoning behind the creation of the tag fdr.
A search for fdr is:question gives 24 results, while searching for roosevelt is:question gives 58. Of course, the United States has had two presidents name 'Roosevelt', so in this case, the tag fdr will - if used correctly - have the additional benefit of helping to distinguish questions about FDR from questions about Teddy Roosevelt.
In fact, right now, only 4 questions are tagged fdr.
Looking at the lists produced by the searches above, there should almost certainly be more. However, what is also clear from those search results is that many of the questions that mention FDR are not actually about FDR (for example What was the economic impact of WW1 on USA's economy?), and so should absolutely not be tagged fdr.
In the case of stalin, the situation is slightly simpler. It is unlikely that we will be getting too many questions about 'Fred Stalin from Cleethorpes', so the only purpose of the tag here is to distinguish between questions about Stalin and questions that just mention Stalin.
We currently have 50 questions that are tagged stalin, while a search for `stalin is:question' gives 151 results. Clearly this tag is needed. But it only has value if it is used consistently and correctly. Looking at the lists, it seems this has not always been done.
For example (in my opinion), the question Why did the rest of the Eastern Bloc not invade Yugoslavia? has been tagged stalin when it should not have been, while Were there any attempts to assassinate Joseph Stalin? has not been tagged stalin and it really should have been.
Finally we come to churchill.
I couldn't find a meta post about the creation of this tag, and it was created before I joined, but my first question on seeing that tag would be "Is that Winston, Jack, or Sarah?" (in much the same way as, if we ever get enough questions to justify the creation of a tag for Oliver Cromwell (the Lord Protector of England), we name it in such a way as to avoid the possibility of confusion with Thomas Cromwell, or even Richard Cromwell).
I think this illustrates the value in gaining input from the community before creating new tags.
We have 17 questions currently tagged churchill, while a search for churchill is:question gives 72 results. Once again, the tag does not appear to have been used consistently or correctly in all cases. The question Which peace offer did Hitler give to Great Britain in 1940? has been tagged churchill, and probably should not be. On the other hand, the question Is it true that Winston Churchill said 'I bet my head on Crete' (or a similar statement)? is not tagged churchill, and it probably should be.
Now, I am not proposing a single massive re-tagging operation across all the questions and tags on the site. That would be hugely disruptive. But if people really want to improve tagging on the site, just correcting a few of these each day, to ensure that tags are only used where the question is about the topic of the tag, it could eventually give us a system of tags that is genuinely useful.
Of course, this highlights a good reason for what you have called "minimalism in tagging". Fewer tags require less maintenance, and less maintenance means less disruption to the site. The burden of that maintenance falls on the community as a whole, and the impact of that maintenance affects us all.
To give just one further example.
In another meta question you proposed the creation of the tag east-india-company for the 'Honourable East India Company' (HEIC). A search for 'east india company is:question' currently gives 31 results, and a brief check of the list shows that many of those questions mention the company, but are not about the company (for example Why has Nepal never been conquered or colonized?).
Similarly, some of those questions are about the VoC (Vereenigde Oostindische Compagnie, or '* Dutch East India Company*'), although probably not yet enough to justify creating their own tag (searching for voc is:question returned 3 questions, while searching `dutch east india company is:question' returned 5 questions. Interestingly, only 2 of the questions from the first search appear in the second. Those 2 are actually questions about the VoC, while the 3rd isn't).
It is worth noting that the HEIC also appears in answers to questions which don't themselves mention it. An example here would be: What percentage of the British Empire's economy was profits from the opium trade?. (Clearly, this question isn't about the HEIC, and so shouldn't be tagged, if and when we do agree that the tag should be created.)
The point here is that it is possible that an answer to a question will occasionally show that the question actually was about the tagged topic, even if the OP of the question didn't realise it, and the question didn't mention it. In that case, it is probably appropriate to re-tag the question (ideally, perhaps, when posting the answer). I did scroll through the list of answers and didn't see any in this case, but it is always worth checking.
So, given all the above, in my opinion, it probably is worth creating a tag for the HEIC. The value of the tag would be disambiguation between questions about the HEIC and questions that simply mention the HEIC. However, we should be careful when choosing its name, and should also probably create relevant synonyms and an appropriate description for the tag at the same time.
But, to be absolutely clear, that tag only has real value if it used solely for questions about the HEIC, and not just for any question that happens to mention the company.
It already works that way for me. With all due respect, you're not the entire SE user base. Proposing changes to site policies has to gain the support of the community (typically, by explaining the advantages); you cannot demand the rules change to accommodate you.
– Semaphore♦
May 6 '20 at 11:01
While in broad terms I agree with Sempaiscuba's answer, it looks to me that it highlights on part of the Help Centre document. To me, there's an equally valid other part there:
A tag is a word or phrase that describes the topic of the question. Tags are a means of connecting experts with questions they will be able to answer by sorting questions into specific, well-defined categories.
Tags can also be used to help you identify questions that are interesting or relevant to you.
I don't think we should always tag to five tags, but I think that the tags should give an idea of what the question is about -- in both time and space as a minimum. In some of these cases, this might mean five tags, in some it would be one or two -- some subjects cover several topics and those will obviously need more tags.
But, to get closer to my point, I think that the sempaiscuba answer doesn't highlight how tags are meant to help us find the questions we should know most about and be able to contribute most in.
There certainly are tags at present that I don't think should exist -- cases in point are actually the fdr, churchill, and stalin that the above answer brings up. The problem for me is that while there may be numerous questions about these characters, the underlying principles behind these people (and, yes, John should be before Winston!) don't change -- they formed the political leadership of their country. But, all of them also had their trusted lieutenants without whom they would not have been as effective.
As it doesn't make sense to make a tag for every general that one can ask about -- there would be hundreds of low use ones that would both clutter the navigation and make subscriptions annoying (imagine receiving an email a week for every WW2 general...). Perhaps -- and in this case this is presumptive only because this retagging would be incredibly annoying -- there should be a political-leadership and if we use usa, world-war-2, and political-leadership together in one question, it's clear that we are interested the specific period, country, and set of people. This would be more useful for the UK where -- perhaps half? -- of questions relating to Winston Churchill might actually involve things executed by Attlee or Halifax.
On the question of East India Companies, I don't think there should be a tag for the British East India Company over the Dutch (or any other) one. There should be a uniform one that applies to all chartered trading companies (as in the topic in question) such that they are all grouped together -- hence, chartered-companies works very well. This makes sense because if someone is on the tag watch, they can subscribe and see questions on the Portuguese, the Dutch, and the British. Also, companies which engaged in trade on the same principles in Canada and the US should be included -- the HBC doesn't need its own tag, etc, because it would be too low-use while not offering a significant contribution.
Yet, the person interested in the HEIC/VOIC is also likely to be interested in the operation of the HBC, if not due to the time and location, but due to the nature of the development of a commercial company and how its rights progressed. If this doesn't stem to "interest" (maybe they got taken around a few too many malls as a child), I think it still indicates a useful link to topics that the person would have knowledge of. This should also lead to answers with that were perhaps not intended, but which are better for referencing other similar entities and increasing our understanding of how the entire system worked by bringing the question to the attention of the experts in that field.
It was on similar principles that I previously said that a christian-military-orders should exist to cover Calatrava, Sword Brethern, etc. I think I was a bit mistaken in that because it would actually make sense to use this general tag for the Templars, Hospitallers, Germans, Calatrava, and everyone else. Same boot. Once we have 25 or more (arbitrary number this '25') questions on an Order also give that a separate tag, to be used alongside. So, in this case, we would have one type-tag (much like trade or diplomacy) which people could subscribe to in order to learn about the feature, along with more specialized country-tags (what they were in effect) to focus on those.
With regards to tags being specified on questions which don't belong there, much like what sempaiscuba described above, there should be a slow but steady cull. The only way this can work, however, is if the tags have a description to describe what they should be for. Even if the name of the tag seems to fit, the purpose behind that tag can only be described by the tag wiki.
Nevertheless, we should strive to remove (please note this is quite different from the very separate and far more difficult question of 'When should other suitable tags be added to a question?') the unsuitable tags from questions when these are apparent (say the examples from above where fdr is used for things that don't relate to FDR).
I didn't mean the notes above on fdr, stalin, etc as actual propositions that those should be deleted. I was merely indicating the way that I think tagging should work: a method of categorization to help people subscribe to topics they like and to find questions they might know about. Obviously cases will extend outside of this and that doesn't mean that other more specific ones might not be necessary: perhaps the same stalin being a case in point.
Finally, as the study of history is generally focussed in time and space -- not many people studying the entire history of everywhere -- all questions should be placed into a geographical location and time from the tags associated with that question.
