In 2014 I noticed a number of questions with version tags, but the question or answers are not version specific.

Having the version in the question body is certainly useful and appropriate for all questions, but I think also having it question titles and tags is another story, unless it truly is version specific.

This was mentioned in comments when version tags were discussed in 2010 at Formatting tags relating to version numbers? Consensus appeared to be that version tags should only be used if the question is version specific.

I wholeheartedly agree - to tag just because that's the version being used, even if it's a generic question, devalues the version tag when specifics are used to narrow search results (more on this below). In 2013 the subject was also discussed in terms of retagging and editing old posts when software went 'out of date' at Should tags be removed because they refer to old software? where again consensus appeared to be only if version specific. The subject was also touched on specific to ArcGIS at Tagging ArcGIS questions?

What I am asking about / proposing is not about going back and retagging lots of old answers. What I am interested in is recent questions and moving forward. We seem to have forgotten the original consensus, or at least not be implementing it. My first idea was this question, to bring it back to people's minds to be aware of as they edit and post.

Or perhaps the original consenus no longer holds. A point raised by @PolyGeo below is that use of version tags has shifted from earlier consensus (which seems to have held some sway into 2013 based on comments in the above cited questions). The 'newer' approach and use of a version tag is perfectly valid and reasonable – quickly narrowing down questions to consider based on what you have access to and getting an idea of user base.

But to my mind this is at odds with the earlier approach - if everything is tagged a version just because that's what asker is using, then the tag means nothing in terms of whether the question or issue actually has any relevance to version number. If I browse the 10.2 tag right now, most of what I see are questions that could be solved the exact same way in any version of ArcGIS, and so an tag means the same thing as the generic tag.

That also hinders searching in a couple of ways. A searcher may not even consider the result since they don’t have that version. Conversely, often only the version tag is applied and not the parent software, so general issue search results may not present a version result that has an applicable solution. And with a generic question, any of multiple results would suffice but a version tag could help narrow results, particularly if there is a difference that depends on version. As has already been pointed out, it may not be possible to know if a question is version specific without an answer.

One solution to the version tag only problem is present in other tagging systems, but I don't know that it is here: if a particular tag is added, its parent tag is also automatically added (ie, adding tag automatically adds tag). Note this is different than synonyms.

I also wonder if we should revise some of the tag wikis so that all version tags abstracts either say or at least start with something like for questions that apply *only* to this specific version of the software (for my use of version, something different for that PolyGeo mentions)? Currently they're varied in format and generic - 'a specific version' seems kind of obvious and doesn't imply any meaning or reason. Some seem to place what the software is into the abstracts rather than direction for use of the tag (granted, appropriateness of that varies by tag).

Can the two approaches to using tags co-exist, or do they contradict each other like I view it? Which way should we move forward, and how do we encourage that?

Examples - why should these questions (grabbed at random with a check running down the list) be tagged ? The question and answers appear to apply to any version of ArcGIS to me and the version is irrelevant. (Again, this is from the viewpoint version tags indicate a specific relevance as opposed to a more generic 'being used'.)

  1. How to auto increment a field in a feature class?
  2. How to generate StreetFrom and StreetTo data from street segments?
  3. Calculating percentiles of a raster (TRMM DATA)

If a function changes between versions, or a tool is new, sure. I usually see that noted in the answer like here. Further, with that question specifying version in body definitely needed. But the question itself, and the solutions not noted to require a different version, are not specific to 9.3.

On the other hand, with this question, if the Split tool has changed where it used to order in sequence and no longer does, the tag would certainly apply. Or 'why does my data frame change projection when I add a basemap' would need (at least) a 10.1 tag because it's a bug in that version.

  • 2
    Version tags triggering product tags, working as you describe it, would have problems tagging an ArcGIS 10.2 for Server Question. – PolyGeo May 8 '14 at 22:08
  • 2
    Can you provide some example questions demonstrating the problem you perceive? I don't see any major problem with tags at the moment. I would also suggest looking to other Stack Exchange sites (e.g. StackOverflow, SuperUser, etc.) to see how they handle version tags. My impression is that they are a necessary evil -- necessary because they provide a way to find and classify questions applicable to a specific version of a product -- but evil because they are semantically redundant (as they usually contain the root tag name of the product). – blah238 May 8 '14 at 23:33
  • 1
    @polygeo When a new version is pushed (ie, 10.2) all Arc products (platform) are updated at the same time? If that's the case, it would mean dropping -desktop from my example in the question. I tend to assume, perhaps wrongly, whenever someone just says a version, they're talking about desktop. I recognize the point, though it reinforces that a version tag by itself doesn't provide sufficient information. – Chris W May 9 '14 at 0:10
  • 2
    @blah238 I don't know that there is a problem now. I was under the impression from the above citations that version tags were only for questions that were specific to that version. That also seems to be the way the other SE sites work (admittedly I haven't dug all that deeply into them). But if I am mistaken then that's on me and there is no issue; I just can't use version tags as I thought. I agree with your necessary evil assessment, but is it "applicable to", or "only applicable to"? – Chris W May 9 '14 at 0:26
  • 1
    When I see a Question tagged arcgis I wonder whether it is Desktop, Server, Online, ArcObjects or something else. Likewise if there is an ArcGIS version tag and no ArcGIS product tag - but at least I know the version being used. I have tried to explain why I think an ArcGIS Question needs both a product and a version tag at meta.gis.stackexchange.com/questions/3131/…. I have tried to be pragmatic rather than purist to try and cater for requirements from various GIS SE users who use the ArcGIS Platform. – PolyGeo May 9 '14 at 8:37
  • 1
    I did further digging on SU & SO last night; seems they have differing approaches. SU has a tag format brand-product-version as a single level tag (Excel is the example that came up most). That's a lot of tags. SO has product (for all) and product-version (only for specifics). There is question or suggestion that if you follow product, do you also see product-version? It would solve problems if so, but doesn't appear to be the case. Implied tagging is certainly not without problems. – Chris W May 9 '14 at 18:16
  • 1
    Looking back at this just now ... where you asked "why are these ... tagged 10.2"? The answer in each case is that their asker tagged them that way and I suspect because he/she was using that dot release and wanted us to be aware of that as we began to think about what might and might not apply to it. – PolyGeo Jul 14 '14 at 4:01
  • 2
    @PolyGeo I'm sure you're probably right, but in the context of my question it was more if/why should they be rather than how they got that way. I may have lost that continuity over the course of edits, so I'll make another to clarify that section. As mentioned in my opening, yes having the version mentioned in the question is important for just that reason (knowing the details), but my stance on tagging was was that a tag should only be applied if the question did truly end up being version specific - something that might not be known initially. – Chris W Jul 14 '14 at 18:23
  • Might not be known initially and in many Q&As is unlikely to be knowable ever because we rely on volunteered effort - see answer by @blah238. – PolyGeo Dec 30 '16 at 21:09

I agree that version numbers in question titles rarely add value, with one possible exception being a highly focussed question like "Is XXX tool broken at Product 8.4 SP9?".

I also agree that the version number(s) of any product(s) being used should almost always be mentioned in the question body. I think versions should be explicitly mentioned in the body even if they have been included in the title (which is meant to only summarise the body) or as a tag. If the tags contradict the version mentioned in the body - I trust the body.

Where I differ is that I think version tags maximise their value when used to indicate the version being used rather than having a rarely practical requirement to be the only version to which a question is relevant.

The implication of suggesting version tags are used "for questions that apply only to this specific version of the software" is to me that they would very rarely be able to meet that criterion. Under, this scenario, if I notice a problem in version M.N and suspect that it is only in this specific version, then due diligence would mean testing it in versions M.N-1 (the version before) and M.N+1 (the version after) to show that it worked in both of them, and is only broken in the version in between. Having multiple versions available to do this is only likely in larger GIS shops, and time to do it may be in short supply. Only then could I meet the "consensus" criterion reached a few days into the six and a half year history of GIS SE and "legitimately" apply such a tag.

Consensus is easier to claim than to demonstrate, so I will make two statements, and hope they resonate. I use ArcGIS platform questions as my examples purely because those are the ones I am the most familiar with.

  • There are often quite major differences between even dot releases of ArcGIS products, and usually quite minor differences between service packs. Consequently, we seem to have settled on version tags like , , , etc and service pack tags like are no longer present/used. We seem to have agreed that this is a suitable level of granularity.
  • Version tags, when used, indicate the version that the asker was actually using and have not for a long time indicated that the issue being discussed is only present in that version.

Assuming that this description of how version tags have been applied for at least the past few years (I think 5-6 years) is accurate, then we can look at how such version tags are used by searchers and potential answerers:

  • Searchers, such as the regulars likely to be reading this, may use version tags to find answers, but I suspect the vast majority of searchers are occasional visitors to this site and that they are all but oblivious to tags placed on questions.
  • Potential answerers, on the other hand, are likely to be looking for that elusive subset of all questions that are likely to match their skillset and available tools so that they can answer them quickly.

As a potential answerer, I can say that when I am looking for questions to answer I first use the tag because that is the ArcGIS product I have licensed, and with which I am the most familiar. The second tag I reach for is because that is the version of ArcGIS Desktop that I currently have installed. Knowing that every question tagged and has been penned by someone using the same configuration that I am using gives me hope that I will be able to test and respond to what they are asking.

When I get through my "list" I start on because, in terms of ArcGIS for Desktop, there is little difference between those versions. After that I work through , down to , but I do not recall searching for or for many years.

One more piece of value that I ascribe to version tags when used with ArcGIS platform products is the market intelligence that they enable to be gleaned. When I am thinking about what version of ArcGIS for Desktop I should be focussing my skills on there are not many ways to get an estimate of what version the ArcGIS user base is predominantly using. I could ask an Esri office for figures about what percentage of users are at 10.5 vs 10.4 vs 10.3 vs 10.2 vs 10.1 vs 10.0 but the likely answer is that "almost everyone has upgraded, and those that have not are about to ... (so you should pay your maintenance)". Instead I turn to GIS Stack Exchange which now has a large number of questions that I can mine for the breakdown of people using each over different time periods. By doing this I feel far more confident when I decide to "turn off" my learnings and services on an older version in order to focus more on the version(s) my potential clients are actually using.

  • 1
    It seems then perhaps consensus has not been reached, or no longer holds. :) I've revised my question based on your answer. I agree with you about title/high focus. And I don't think we differ on version tags being useful/valuable, just how or to what purpose. I might also clarify that by consensus I just mean among those who discussed or posted comments specifically addressing use of version tags. I didn't see (might have missed!) any dissenting views about how the tag should be used in the cited questions, only when you pointed it out in this answer. – Chris W May 8 '14 at 20:57
  • 1
    @ChrisW I think ideas on how version tags should be used were only ever floated briefly and any consensus was only reached via brief common usage rather than discussion. I'm glad you asked your Question because it has given opportunity to document some of the pros and cons of the divergent ways version tags can be applied. – PolyGeo May 11 '14 at 7:14

I agree with @PolyGeo's answer to this but just wanted to add a couple of points:

  • For many users it's difficult to actually know whether a question or problem is exclusive to a particular version but they usually know the version that they are using.

    Since it's better to err on the safe side and include rather than exclude information here, the better default is to include the version tag, for that and the other reasons already expressed.

  • It is possible use wildcards to search by tags. E.g. searching for [arcgis*] automatically expands to the search:

    [arcgis-desktop] or [arcgis-10.0] or [arcgis] or [arcgis-10.1] or [arcgis-server] or [arcgis-javascript-api] or [arcgis-9.3] or [arcgis-10.2] or [arcgis-engine] or [arcgis-flex-api] or [arcgis-online] or [arcgis-silverlight-api] or [arcgis-rest-api] or [arcgis-runtime] or [arcpy] or [arcgis-runtime-sdk-wpf] or [arcgis-android] or [arcgis-mobile] or [arcgis-platform] or [arcgis-flex-viewer]
    
  • I have seen conflicting guidance on other meta sites about this topic, but with the passage of time I think it has become clear that using version-specific tags has become the norm across SE, with the more technical sites such as ServerFault especially favoring them because in that realm, versions really do matter a lot -- as they do in ours, I would suggest.

    On the other hand, sometimes version numbers can be overkill, as was decided over on SuperUser regarding the Firefox version-specific tags.

  • 1
    I agree it's difficult to know, as mentioned in the question. We're a little different than SO, where if you're programming for or using features in a particular version it's pretty clear up front. Didn't know about the tag search, and I'll keep an eye on that. Does it work for tag subscriptions? Also agree on conflicting guidance - it seems to be up to the particular community, and we seem to have changed our minds. And I saw the Firefox example, and wonder if we were to check version tagged questions in how many would version actually be relevant or if they'd just go away as in that case. – Chris W May 9 '14 at 22:09
  • 1
    I agree with @blah238. In towards 100% of cases the asker knows their version (and potential answerers frequently ask what it is when absent - because it helps them know their answering options). In towards 0% of cases does asker know if issue is related to single version only, and with latest version that only becomes knowable posthumously. Tags help answerers far more than they help searchers. We seem to have been working to a very usable model for several years so let's not resurrect the first way we tried on version tags. – PolyGeo May 9 '14 at 22:51

I fully agree with the arguments you make in the question, which I consider an answer in itself.

Version tags should only be used when answers to a question will only be useful to that specific version. Matt Wilkie makes a nice summary in his comment on Formatting tags relating to version numbers?

People should be encouraged to use version tags only if there really is something version specific about their question, and it's perfectly okay for someone other than the poster to change a tag if it really isn't version specific.

In my opinion, the alternative approach suggested by @PolyGeo goes against the system design, let's see:

All the other SE communities I currently participate, follow the approach suggested by Jeff Attwood's (one of SE's founder) in What are the guidelines for using version-specific tags?, which is in line with arguments made in your question and in this answer:

In general, you should strongly avoid creating version tags.

I would only use version tags when the question content is irrevocably tied to a specific version of something and can never be relevant to earlier versions or later versions.

In other words, it is perfectly sufficient to tag your question when it contains c# code. You do not need to say that it is because it uses LINQ features. c# is c#!

The only questions that should use the tag, in my opinion, are those questions which are specifically asking for detail about features that were added in 4.0.

  • Dot point 1 may appear like a superfluous edit but it was applied to an ArcGIS Desktop with PostgreSQL question where functionality has expanded greatly at 10.x dot releases so knowing the version was important. Dot point 2 suggests splitting content when my advocacy is always for appropriate granularity and in this case I would be merging content (questions). Dot point 3 is about the main purpose of tags which I think is to help potential answerers find the subset of questions they want to work on. – PolyGeo Dec 30 '16 at 20:38
  • I agree with the Stack Exchange views of @JeffAtwood near 100% of the time but that 5.5 year old answer is one place where I do not. I wonder whether that is a position that he still holds because over the years he has been known to revise some of his positions, like we are all sensible to do as our understandings of issues evolve. – PolyGeo Dec 30 '16 at 21:22
  • That post was not specific to 10.3 version and the version already was included in the body, hence, the edit did not add any new value/information (dot 1). The post is tagged 10.3; users using 10.2 or 10.4 versions might refrain posting answers there (dot 2). That tag is irrelevant for "searching", in that context (dot 3). @PolyGeo – Andre Silva Dec 30 '16 at 21:38
  • About JeffAtwood's answer, no problem, but it is there and I think it has hold until today. Your proposal is different from what most SE sites do. If the community wants this way, there is no problem, but the decision should come from the community (so let's wait some time to see where this answer can go). – Andre Silva Dec 30 '16 at 21:51
  • That 10.3 tag is useful to me and I assume other potential answerers of ArcGIS Desktop questions. If I currently have 10.3 installed it lets me find questions from users who are using the same version via a single-click or filter. I'm actually using 10.5 but once I review 10.5 and 10.4 questions I review 10.3 next. – PolyGeo Dec 30 '16 at 22:18
  • As useful as version tags can be when applied consistently, they are nowhere near as important as product tags to get "right". – PolyGeo Dec 30 '16 at 22:25
  • @PolyGeo, so apart from our disagreement about how version tags should be used, do you agree in that example, the arcgis-desktop would fit better than arcgis-10.3? – Andre Silva Dec 30 '16 at 22:29
  • If four important tags were on the question and I had to choose between a product and a version tag for the fifth tag then I would choose the former. If there was room for two tags and I knew the product and version being used then I would include both. – PolyGeo Dec 30 '16 at 22:53

You must log in to answer this question.

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