4

This is a follow up inspired by Improving our tagging (redux). Let's think about tags for dates and date ranges.

I'll begin by listing some of the prior questions that may bear on this question:

Here's a thought-experiement I would like everyone to think about, for this question and for any of the other re-tagging questions. Imagine that we are at a conference like RootsTech or Who Do You Think You Are Live, and we have volunteered to work a room where people can come to ask us questions. We have ribbons on our name badges and a sign behind us where we can pin up tags so the people asking questions can see which desk to go to, depending on what kind of question they have.

Take a look at your profile and sort your badges into classes. Pull out a list of your tag badges. What do you see?

I have one silver tag badge for , and these bronze badges:

enter image description here

You can see at a glance that I am interested in methodology, that most of my answers have been about the United States and England, and that my focus so far has been in the 19th and 20th centuries. It shouldn't be much of surprise to see that I've also answered a lot of questions about census records; for England and the US, the most useful census records are (surprise) in the 19th and 20th centuries.

What do your tag badges tell you about your answers and your self-answered questions? Are they an accurate reflection of what you are interested in answering, or that you are likely to be able to answer?

Newcomers aren't likely to look at our profiles to see which badges we have. But it does give us a way to look at the tags as Jeff Atwood suggested on this post on the Stack Overflow blog:

Periodically asking questions like “who would ever subscribe to this tag, and why?” can reveal a lot about the nature of tagging on your site.

If I were a person in a hurry, and the site was busy enough that I only had time to look at these tags (either by search or subscription), then would this give me a selection of questions I could likely answer? Yes, it would. I'm far more likely at this point to be able to answer questions about the 20th and 19th centuries than I would about the 18th, 17th, or earlier.

Now -- for breaking down questions into time periods, is this good enough? Is there another way, and if so, is it a better way?

One could divide questions in England into before-and-after 1837, because of civil registration and census records after 1837 and (mostly) the lack of them before.

Canadian questions could be before-and-after Canadian Confederation.

For US Census records, there's a big divide before-and-after the 1850 Census because of the lack of every-name census records. For immigration, pre-and-post 1840, for naturalization, pre-and-post 1906. In the USA also, the colonial period is a different kettle of fish than later on when you have federal records.

Would it buy us anything to use tags to sort out questions from the different eras?

Maybe your answer is "no" and the century tags are enough. But what I'd like us to think about are ways to group the site's questions into pools that a person could look over to get inspiration for their own question, and that those of us who have 'expertise' (in the loose sense, not the I'm-certified sense) in an area could subscribe to a tag in case they needed to -- or to point a new user to a group of questions that might be useful.

In the same way we might say "you have a DNA question, why not look at the other questions tagged ", are the century tags enough to do the job?


Now that I've seen some of the answers, I wonder: do we need time period tags (specifically, the decade and century tags) at all?

2
  • 1
    yes to century, not to decade -- for reasons I've already outlined. – user104 Mar 31 '17 at 16:49
  • Jan, do you think we've reached a conclusion on this and are ready to move on to discussing what makes a good tag wiki for a time period tag (and in parallel perhaps discuss the uses of geographic tags?) – user104 Apr 22 '17 at 8:46
3

I see three classes of 'time period' tags that we need to discuss, and

  1. want to assess each of them in the context of what is needed of a tag
  2. talk a little about an alternative approach that JanMurphy raised.

What is needed from a tag?

In various places, we are told:

  • Tags are a map of what our community is about/not about (and so implicitly about allowed and disallowed topics)

  • Tags are for organising (not summarising) content, to group questions into specific well-defined categories, so that:

    • experts can find suitable questions they might answer
    • those looking for answers can find potentially-related material that might address their need
    • users can filter their view of the site to highlight material of particular interest/downplay material of little or no interest
  • The most-popular (or only) tag on a question is used for SEO purposes (prepended to the page title in search engine results) -- worth noting, this can reduce the amount of a question title that is visible significantly for long tag names.

  • Tags must work as the only tag on a question.

(All this is irrespective of how people choose to be alerted to new content -- newsletters, rss feed or whatever-- the medium doesn't affect the purpose of tags).

Classes of 'time period' tags

The three classes of 'time period' tags I suggest we discuss are:

  1. 'Named' time-periods e.g. , ,
  2. 'Century' tags e.g. ,
  3. 'Decade' tags e.g. ,

For comparison purposes, history.stackexchange.com make much more use or Class 1 (Named periods) than we do. They use century tags, but most of them are not high up on the list of popular tags (which mitigates some of their disadvantages). They use decade tags very sparely, and only for the 20th century (I'm guessing there are fewer widely-recognised 'Named periods' in the 20th-century.)

Assessment of current time-period tagging aproach

1. 'Named' time-periods e.g. , ,

We use these for questions related to events that took place in a widely-recognised named period, and/or records that were created specifically during such periods.

Allowed/disallowed topics?

Yes (except the absence of a tag for a particular named period doesn't make it automatically off-topic, just not asked about yet.)

Useful to organise material?

Yes (and there doesn't seem to be tendency to use them to summarise rather than organise).

Meaningful and helpful as the first element of a Search engine result?

Yes, because of the degree of specificity involved.

Works as a lone tag in a Genealogy context?

No. Usually a geographical tag is necessary as well, as well as a more a genealogy-specific tag, but that applies to most of our questions.

Conclusion

'Named time-period' tags (IMO) are good, useful specific tags and should remain. And we could extend this class of tags when necessary (e.g. , ).


2. 'Century' tags e.g. ,

We use these for questions about events and/or records created in the relevant century, or people who lived in that century.

Allowed/disallowed topics?

Not really helpful, as any century for which there are genealogical records available is on topic. Plus, they're based on a western-centric calendar and we don't want to discourage questions based on other calendars (hoping the relevant experts will show up to answer them).

Useful to organise material?

Possibly, although the availability of genealogical records in any particular place rarely aligns with a century boundary

They might be useful as a very-high-level categorisation so that people can filter out noise about centuries that aren't relevant to their interests/expertise. However, there's a tendency to add a century-tag to every question, because they exist, even if it's just summarising the question (not always accurately) rather than categorising it.

Meaningful/helpful in Search engine results?

I say not. They're likely to be among the most popular tags (because they're ubiquitous) so will end up at the front of many results and displace a more relevant tag; even though they're rarely the most important thing about a question.

Also, the time period in question is almost certain to be specified in the title or body of the question and so affect the search results.

Works as a lone tag in a Genealogy context?

No. Usually a geographical tag is necessary as well, as well as a more a genealogy-specific tag, but that applies to most of our questions.

Conclusion

I'm in two minds about 'Century' tags.

There's a tendency to use them because they're there and 'must therefore be useful' but this leads to a number of drawbacks (especially in search engine results).

They could have some utility for high-level filtering. Wen our traffic volume goes up, I'll probably concentrate on 18th, 19th and 20th century in England and Wales, for example -- it's a high-level filter but will cut out most of what I can't help with. But then so would just filtering by England and Wales... But I recognise the same approach might not be true for other geographies -- an American genealogist whose ancestors immigrated in the nineteenth century may not be interested in sixteenth and seventeenth century colonists and immigrants, or their records.

IMO, we wouldn't suffer if we significantly reduced the use of century tags as long as we used other more 'intelligent' tagging approaches, but I'm not sure we're ready for the re-tagging workload that would involve; and there are cases where they are an appropriate distinction. We should certainly discourage using a Named-Period and Century-Period tag on the same question -- one makes the other redundant. (Note: using a mixture of 'Century' and 'Named' tags will potentially increase the number of tags an individual will need to monitor.)


3. 'Decade' tags e.g. ,

We use these for questions events and/or records created in the relevant decade, or people who lived in that decade.

Allowed/disallowed topics?

Not really helpful, for the same reasons that century tags fail this test.

Useful to organise material?

Much less so even than century tags, and for the same reasons.

In particular the range of expertise of an individual is very unlikely to be restricted to a single (or small group of) decade(s).

It's been said that (multiple) decade tags help narrow down searches, but that only works for 2 or three decades. If a question spans 4 decades or more it may not show up in your search because there isn't room for 4 decade tags plus a geography tag and a tag that's really useful to categorise the question! And if search results aren't consistently useful, they're no use at all. (If you really want to search on a specific date range mentioned in a question, Google is a much better way of doing this, and it brings up answers as well as questions).

Meaningful/helpful in Search engine results? No. This is almost never the most important thing about a question, but again they're ubiquitous.

Works as a lone tag in a Genealogy context?

No, for the same reasons as all other time-period tags.

Conclusion

As you can probably guess, I think we should ditch 'decade' tags, as they serve no good purpose that can't be met in another way, thus freeing up 'tagroom' for more useful categorisation (and possibly even making people think more carefully about the right tags to use). And if they are retained, they should be used instead of the other time-period tags, not alongside them. And we should be vigilant to retag questions where another period-tag is more useful.

Is there a better way?

To address Jan's questions about 'Is there a better way' [rather than using calendar-based time periods such as centuries]?

Well, yes, sometimes, but it would involve a lot of work (including -- ta-da -- creating new tags, improving our tag-wiki and re-tagging questions) but IMO the payback would be worth it. (and better to do it now than later when the workload would be worse).

An example of where I think a calendar-based tag isn't a good solution, even if the question relates to a specific period, is:

How did the information gathered for civil registration vary with time in England and Wales.

Right now, to categorise that question you'd need the following tags:

and (as civil registration was the same across both countries). You can't use because things differed in N. Ireland and Scotland (as well as sundry islands around these shores).

, and to cover the time span of interest (recognising that the first 37-and-a-bit years of the 19th century is not relevant).

because it includes all of and . (You might argue that the question should be split by record type, but that would involve an awful lot of repetition in the answers).

That's six tags, so which do I lose? Geography is essential. Vital records is essential. So I have to ditch all the time periods as just losing one of them would be misleading... And then people following by time periods would perhaps never see my question.

I think a much better solution would be to tag it . Those three tags (combined with a good tag wiki that specifically identifies the period during which Civil Registration has taken place in England and Wales) categorise the question completely.

Or even: . That would allow the tag wiki to be very specific, but it is a long tag which would affect search engine results. However, could get very unwieldy if it had to cover all geographies, which would make it much less likely to be useful (or even to get created). And of course this approach could be generalised.

But sometimes a time period is the right thing to use, even with a specific record type.

What information would I find in a Wiltshire parish register baptism entry for the early 19th century?

I'd tag that with:

-- no need for as is more specific.

( there might be local record keeping practices) -- it might get retagged to include as well and possibly lose if an expert deemed though the more general tag was applicable instead (i.e. there were no local peculiarities of note).

-- broad time classification. There is a distinction in the records pre- and post-1812, but a good answer would include that anyway, and somebody asking the question may not know that break-point.

And sometimes we can't be specific about record-type either:

Looking for the birth-date of Fred Bloggs in England around 1865?

Thing-place-time again:

.

In summary

What we have right now is a reductionist approach to tagging for time-period (century/decade building blocks) combined with the sparse use of more 'intelligent' Named-period tags. This severely reduces the tagroom available for more important tags and (as others have pointed out) leads to redundant and/or misleading tagging as well as poor search-engine presentation.

What we should have is a simplified set of basic building blocks for use when nothing more specific is relevant (centuries and other major calendar divisions relevant to non-Western calendars); plus an increased set of Named-period tags (created as they're required) plus more specific record-type tags that include geography explicitly and time-period implicitly.

12
  • 2
    I think named period tags could be quite beneficial, for certain locations. For example, the "Periods in English History" tabulated on Wikipedia align fairly well with major social and record-keeping changes (e.g. civil registration started with Victoria's ascension). But a full set of such tags, per country, could be rather burdensome, and I'm not sure how to limit them properly. If each monarch/president/PM's reign becomes a "period" then there's a problem, but sensible and intuitive combination of those reigns isn't obviously easy. – AndyW Mar 24 '17 at 10:41
  • @AndyW, I agree deciding on and implementing a complete set would be hard -- but we don't need to create them until they're needed. They would need to be fairly recognisable so it's easy to decide which to use, but (in the UK) Georgian/Victorian/Edwardian might be a good start, in addition to WW1 and WW2 and 'Interwar period'. (As an aside we don't need to reinvent any wheels if there's a good set of nomenclature we can borrow from elsewhere). 20th century might be the best we could do for 1945-2000 -- or for Qs where no more precision is needed. – user104 Mar 24 '17 at 10:58
  • 1
    @Andy I like the theory of period tags, but also think that it makes tagging questions difficult for inexperienced users. For example, one would likely have to lookup that 1717 was in Georgian England, meaning an extra step in the tagging process that I suspect most people wouldn't want to bother with. The bigger problem with having date tags that are intrinsically tied to certain geographies is that we get a lot of questions about immigrant ancestors such that a question may pertain to multiple overlapping periods. I can see it getting very messy very quickly. – Harry V. Mar 25 '17 at 0:03
  • @HarryVervet I agree about helping in inexperienced users (so the tag wikis would have to be very clear). And I'd have no problem with sticking with century tags for (e.g.) emigrations that started in Victorian Britian and ended in America (where I suspect 'Victorian' didn't apply.) – user104 Mar 25 '17 at 9:48
  • @ColeValleyGirl I think it would be useful to get an idea of what this would practically mean. The 4 most commonly tagged countries are USA, England, Germany, Poland. Maybe we could start by generating era tags for these 4 countries to see better what it would look like. Perhaps you or Andy could submit the "era" proposal as a separate answer – this answer ends up proposing several ideas which makes it difficult to vote on if one agrees with part but not all of it. – Harry V. Mar 25 '17 at 10:49
  • @HarryVervet this answer proposes "What we should have is a simplified set of basic building blocks for use when nothing more specific is relevant (centuries and other major calendar divisions relevant to non-Western calendars); plus an increased set of Named-period tags (created as they're required) plus more specific record-type tags that include geography explicitly and time-period implicitly." Should I make the summary clearer? – user104 Mar 25 '17 at 11:14
  • I thunk it's important to look as the topic as a whole rather than pick elements out of it that might not combine sensibly. – user104 Mar 25 '17 at 11:14
  • 1
    If you think one element is OK in isolation (or disagree violently with one) you can pull that out as a separate answer -- as @PolyGeo has done. – user104 Mar 25 '17 at 11:18
  • @ColeValleyGirl I have upvoted this as I agree that decade tags are not useful, and largely agree with your summary. However I question the practicality of the named-period tags you suggest, if the century tags will also exist. How would one know whether to tag a question victorian or 19th-century? I feel like this is running into the same problem as with the decade tags. It's getting too complex, where we have some questions tagged decades and some centuries. Hence my desire to not upvote this. – Harry V. Mar 25 '17 at 16:01
  • 1
    I also think it is important to distinguish event or conflict tags (such as world-war-1) from era tags (such as victorian). The latter is purely a way to group questions by date, while the former is more than that. – Harry V. Mar 25 '17 at 16:09
  • @HarryVervet I agree about distinctions. I'd use world-war-1 for records/events specifically associated with the war (service records, war graves, pow camps, death in action...) and 20th-century for records that are only coincidental with the period 1914-18 such as births, marriages etc. But I wouldn't use both on the same question. – user104 Mar 25 '17 at 16:16
  • On victorian versus 19th-century, it isn't even just a choice between them -- chuck in georgian as well. So if we had Victorian and Georgian (for UK-related questions only, probably) it's a choice of one of those or 19th-century if a Q spans both periods... but I won't die in a ditch for Victorian and Edwardian if 19th-century is good enough. – user104 Mar 25 '17 at 16:22
2

Say I have a question about an ancestor who served in the Second World War. Do I tag , , , , all of the above?

It will become unsustainable before long. We'll have some questions tagged with decades, some with centuries, some with neither. And that's not helpful for anyone looking for questions about those specific periods.

Here I propose removing the decade tags entirely. My reasoning:

  • People don't know when to tag decades, and when to tag centuries. I don't know sometimes, even though I wrote the usage guidance for many of them. We subsequently have a mishmash of some questions tagged decades and some centuries.
  • Some questions end up having 2 or more decade tags, which in some cases consumes valuable tagging real estate. Or the ones that should be tagged 2 decades only get one because there are more important tags.
  • Any tag that has a decade tag could equally validly have a century tag. For example, with the current usage guidances there is no reason a question shouldn't also have the tag – other than that would be plain silly.
  • Decade tags usually serve minimal useful purpose in grouping questions. Questions rarely are specific to one decade. Near the top of the question list is: Meaning of “dofhoadayne” in 1648 Will of William Noake from Longburton, Dorset, England?. This question is not specific to the 1640s, but it is tagged as such. Palaeographically if makes little difference whether the question pertains to the 1640s or 1670s; hence the usefulness of a century tag which encompasses this entire period.
  • To search for all questions in a multi-decade period, one has to do complex searching techniques (which I suspect the vast majority of users do not know how to do). Tags should cater to the moderately experienced user, rather than those who know the intricacies of the StackExchange search algorithms.

I support the tagging of events or conflicts in addition to the century, because the conflict tag conveys much more than just the time period.

This being said, in an ideal world I think it would be useful to be able to tag certain large record sets that are specific to a year or time period. For example, I think it could be nice to have a tag for . Not for questions that mention the 1881 census, but for questions about the 1881 census. However, at this time I do not support such a proposal because I think it would become a worse tagging mess than the decades. The search function is adequate for findings questions that mention the 1881 census, even though it is not good for weeding out questions about the 1881 census.

The bottom line is that our tagging of dates needs to be simple and straightforward. Century tags are the answer to that. Adding any further tags such as decades or eras just makes tagging more confusing to the average user, which makes the tag groupings less useful to everyone.

7
  • At the moment, the combined search for census-records england 1881 genealogy.stackexchange.com/… yields only 7 results; as you said, it includes questions that only mention the 1881 Census. – Jan Murphy Mar 17 '17 at 1:18
  • If the question is about the second world war, then I would tag it simply as world-war-2, with no need for additional time period tags - we all know that was 1939-45. – PolyGeo Mar 24 '17 at 2:41
  • I agree with the removal of decade tags. Events and administrative changes rarely align on decadal boundaries, and it's not often that a question/answer scope is genuinely restricted to a single decade. Social history has some strong decadal differentiation (particularly in the 20th century), but I'm not sure that family history follows suit. – AndyW Mar 24 '17 at 12:31
  • I would tag it with world-war-2 and no other time period tags. If it was about records (or something else) from the decade leading up to World War 2 then I would tag it 1930s. – PolyGeo Mar 25 '17 at 10:11
  • 1
    @PolyGeo I agree, for many WW2 questions, the world-war-2 tag is likely sufficient in terms of date tagging and would not also require 20th-century. However, I would caution against using world-war-2 to tag questions that are not at all related to the war but happen to pertain to the period 1939-1945. – Harry V. Mar 25 '17 at 10:56
  • Even in Adelaide, far distant from any theatres of war, my understanding is that life was dominated by it. For example, my father raised money for the war effort, my grandfather organized gas masks and sand bags for civil defense, while my great grandfather's car was up on blocks due to the unavailability of petrol. I think most things that I would ask about the period 1939-45 would probably be tagged suitably with world-war-2. However, I'm sure there are some things like my great-grandfather's death in 1943 that I would put a decade tag like 1940s on. – PolyGeo Mar 25 '17 at 11:07
  • 1
    @PolyGeo I don't disagree, if it is a question that is any way war-related, then world-war-2 is fine. Except on that last question you mention would put the 20th-century tag on :P – Harry V. Mar 25 '17 at 16:13
0

OK, I'm going to try to summarise the views I've seen so far after more than a month. (Please feel free to correct me if I've misrepresented the answers so far).

  1. There is general support for tagging by (western-calendar) century -- e.g. . It acts as a (very) coarse filter on questions that -- in conjunction with other tags -- can narrow down the questions that might be relevant to askers and answers (however they seach/otherwise filter questions). It could and should be supplemented with similar coarse-grained filters for other calendars if and when we need them.

  2. There is general support for some tagging by more specific time-period/event/place e.g. where the periods and geographies involved are well understood, as long as the tags are only used for records/events specifically related to the period/event/place concerned (rather than just coincident with that period). An 'event' tag should not be used alongside century-tags as an 'event' tag is more specific. There may be occasions on which we want to extend our 'time-period' tagging, but we need to ensure the periods we pick are widely understood and produced a distinctive set of records.

  3. A more granular approach to time-tagging is possible, that combines time-period/record-set/geography (as per the original question); however, although this would be useful, the effort involved probably isn't justified.

  4. Decade-tags e.g. have little support and should be deprecated. They confuse users, consume valuable tagging real estate, and serve minimal useful purpose in grouping questions. Searching for one or more decades is supported by other means (and searching by tag may be misleading). An argument has been made for using decade tags for the 20th century and beyond, but without much support; we should revisit thisif a good case is made.

If this summary is accepted, I propose we:

  1. Edit the tag-wikis for the 'century' tags to remove reference to the decade tags (but mention that non-western tags can be defined if needed).
  2. Review the tag-wikis for 'period/event' tags to ensure they give guidance on when to use 'period/event' (including relevant geographies) and when to use 'century' instead.
  3. Edit all questions with a decade tag to have a century-tag instead (I'm willing to do take on this workload, as I believe it's less confusing for users than having a tag synonym, althoug that might be an interim step).
  4. Review questions with both 'period' and 'century' tags to decide which is appropriate and edit the tags accordingly.
20
  • The process of merging decade tags into century tags is straightforward and can be done by mods. It would not necessarily require editing of every decade-tagged question. Questions would still need to be reviewed to ensure they are correctly tagged with centuries. However, once decade tags are gone there is no going back, no way to get back that granularity; we just need to be very sure this is wanted. – Harry V. May 2 '17 at 13:55
  • @HarryVervet Or very sure it isn't wanted :) – user104 May 2 '17 at 14:48
  • @HarryVervet I've got enough reputation to suggest tag synonyms (oh for the days when I could just do it :) but that puts a worlkoad on the mods. Great if the mods want to do it, otherwise I'm happy to slog through. – user104 May 2 '17 at 14:55
  • I think removal of decade tags in totality is unwarranted. For recent events, 21st, 20th, and even 19th century I think the granularity that they provide is useful. – PolyGeo May 2 '17 at 20:23
  • @PolyGeo I personally am not dead set against decade tags, but if we keep them I think we should consider getting rid of the century tags. Perhaps that is the answer: if you can't decide which decade to tag, then there's probably no real useful need for a date tag at all. It's the mismash of some questions getting century tags and some questions getting decades that is the most problematic point for me, with no clear usage guidances. – Harry V. May 3 '17 at 0:36
  • @PolyGeo, we all understand you wish to retain decade tags, but you haven't articulated a reason for doing so that has attracted ANY support. Your case seems to rest on using them for searching, but that isn't reliable (because of the limint on the number of tags a question has); nor have I seen any agreement with you that the value of searching by decade-tag (when there are other search techniques that will give more reliable results) outweighs other disadvantages. contd... – user104 May 3 '17 at 9:06
  • . The history.stackexchange website has largely given up on them except for a small number on 20th centurydecade tags that are predominantly used alongside a century tag. – user104 May 3 '17 at 9:06
  • @HarryVervet I think there's sufficient difference between (say) US genealogy in the eighteenth century and nineteenth century (ditto for England and Wales) for century tags to provide an imperfect but useful goupring of questions.I can't say the same for the differences between the 1850 and 1860 decades in those same countries. Perhaps decade tags only would be a perfect solution, but not while we're limited to 5 tags. – user104 May 3 '17 at 9:16
  • I don't want to jump out to Google to perform searching when the granularity of decade tags coupled with century tags for earlier eras allows me to filter using time within G&FH SE. I do not see tags as something that need to be perfect; to me they are simply an aid for potential answerers to find questions that they may be interested in answering. – PolyGeo May 3 '17 at 9:28
  • @PolyGeo "I don't want to jump out to Google..." And I won't want to clutter up questions with decade tags that add litlle value apart from meeting your search preferences. I'd mind less if they actually yielded reliable search results (they don't) and didnt clutter up Goodgle search results that hopefully bring others to this site (they do). – user104 May 3 '17 at 9:40
  • So how do you filter questions on Cornwall in the 1890s within G&FH SE? – PolyGeo May 3 '17 at 9:42
  • You filter on cornwall plus is:question if you want to restrict the results to questions and then search those results for 189*. 5 results. If you filter on cornwall1890s you get zero results and don't have the option to see answers that mightbe relevant. Which is most useful? – user104 May 3 '17 at 9:47
  • @PolyGeo if there weren't other ways to achieve the end you seek, I'd have more sympathy for your case. But sometimes we all have to step back and go: OK, others don't agree with me. I did it on this question genealogy.meta.stackexchange.com/questions/3243/… even though I violently disagree with the majority view. – user104 May 3 '17 at 9:52
  • Both my answers here are at -2 so clearly they do not represent the majority view but that does not make those views invalid to hold. I tried your search and cannot see how this question (for example) relates to Cornwall in the 1890s: genealogy.stackexchange.com/questions/11998/… The case I am making is not about how to search on raw content, it is about how to filter on content curated using tags that are relevant to finding questions (and answers) of interest. – PolyGeo May 3 '17 at 10:39
  • @PolyGeo I guess it depends on whether you'd prefer false positives or false negatives in your result. IMO, it's easier to disregard something you know exists than take into account something you don't know exists. I've never said your views are invalid, just that they don't represent a majority view. And you still havent' convinced me that the difference between two decades is that significant. – user104 May 3 '17 at 10:48
-1

I agree with @ColeValleyGirl that:

The three classes of 'time period' tags I suggest we discuss are:

  1. 'Named' time-periods e.g. , ,
  2. 'Century' tags e.g. ,
  3. 'Decade' tags e.g. ,

but where I diverge is on the use of Century and Decade tags.

I think many questions on this site can benefit from 1 (sometimes 2) time period tag(s), and 1 (sometimes 2) geography tag(s) which leaves 3, 2 or 1 tags available for other purposes.

If a question is tagged with a named time period then I see no need for further time period tags on that question i.e. I think these named time periods are more important to have than centuries or decades.

I started on this site thinking that century tags were more important than decade tags, and that decade tags should not be used. However, I now, after seeing many more Q&As, have come to the conclusion that decade tags should be preferred to century tags, and so, if 1 or 2 decade tags are clearly applicable to a question then I would use them instead of a century tag. If a question covers three or more decades of a century then I would fall back to using the century tag instead.

My reason for this is that I think century tags are too coarse, and particularly so for recent centuries. A tag for every year would be too granular, while a tag for each decade seems to be about right.

For example, I may want to quickly review our questions on East Lothian in the 1820s and 1830s, which I can do by typing [east-lothian] [1820s] [1830s] into the search bar (which currently only finds one Q&A, but imagine a site 10 or 100 times our size).

On the other hand a search on [scotland] [19th-century] in the search bar brings back a far less useful set of Q&As for me.

26
  • Yes but why must there be tags to search for those things? Wouldn't a search for the text in the Q do just as well? – Jan Murphy Mar 24 '17 at 1:04
  • @JanMurphy Tags are used quite differently than search. A question that says "X was born in 1823 and I am looking for where they went just after appearing in the 1881 census" would be usefully tagged 1880s and would not be tagged 1820s. The tag has had a value judgement made by someone familiar with our tagging protocols, and it is that curation that makes this and any curated tags valuable. When I'm looking for questions from 1882-89 (like above) the one-click tag gives many of them to me trusted and instantly. – PolyGeo Mar 24 '17 at 1:46
  • Decade tags violate the "Tags must work as the only tag on a question." criterion. How likely is it that someone will need to ask about how to do genealogy only in a specific decade? It makes sense for missing census years like the 1890 census but there's no other example I can think of, and if people are looking for census substitutes, or substitutes for any missing records, IMHO they are better off being tagged as such. – Jan Murphy Mar 24 '17 at 3:15
  • @JanMurphy "Every tag you use should be able to work, more or less, as the only tag on a question." (with my bolding). In this case its less as the only tag and more as a qualifying tag to something like a geography tag. SE sites have many examples of tags that are only used in concert with other tags to achieve their potential, and the agreement to do that comes out of their community discussions, just like we are discussing here. – PolyGeo Mar 24 '17 at 3:57
  • PolyGeo, how does that differ form the following google search? site:genealogy.stackexchange.com 1880..1889 – user104 Mar 24 '17 at 6:56
  • Also: "If the tag can’t work as the only tag on a question, it’s probably a meta-tag" and "The reason meta-tags are a problem is that they do not describe the content of the question" – user104 Mar 24 '17 at 7:00
  • And finally, the time frame "after 1881" is the least important aspect of your example question, a long way behind the methodology and sources somebody would use (which are almost never decade-sensitive). But the use of an 1881 tag would disgtort the search engine results. – user104 Mar 24 '17 at 7:02
  • "how does that differ form the following google search?" - it's a lot less to type – PolyGeo Mar 24 '17 at 7:37
  • "it’s probably a meta-tag" - in this case it is not a meta-tag although I agree that the "rule of thumb" that you mention can often indicate one. – PolyGeo Mar 24 '17 at 7:38
  • I've not proposed using "an 1881 tag". What I have said is that an 1880s decade tag is useful to me when I am trying to filter questions using that era in combination with other, usually geographic, tags. – PolyGeo Mar 24 '17 at 7:41
  • @Polygeo, I wan't to say something about 'curating' as well, which ties into your comment about meta-tags. Sometimes a decade tag may not be a meta-tag if it was really germane to the question i.e. the answer would depend specifically on the decade in question. But havinf decade tags on the wite wil increase the 'curation' workload that falls on a few individuals -- mostly remaining decade tags that don't shouldn't be there (which will be most places). And it would be hard for newcomers to understand why most often their decade tags were removed... but some were left in place. – user104 Mar 24 '17 at 7:42
  • Re searching, it may be more to type but you can be much more precise. – user104 Mar 24 '17 at 7:44
  • For some purposes more precision can be great; here I use tags as a complement to the SE search methods, because of their convenience, and so it is suitable granularity rather than precision that meets my requirements. – PolyGeo Mar 24 '17 at 7:48
  • 1
    I understand how decade tags are nice in some cases, but the biggest problems in my eyes are that we also have century tags, and the use of decade tags is extremely subjective, both of which precludes these tags from being applied in a consistent and useful way. For example, the question you mention in your search results: genealogy.stackexchange.com/questions/9663. You've tagged it 1820s and 1830s. There is no mention of the 1830s in the question. Why not add 1810s as well, because the answer would be the same? – Harry V. Mar 24 '17 at 10:14
  • 1
    @PolyGeo agreed -- tags can be adjusted before during and after a question is answered. – user104 Mar 27 '17 at 15:45
-1

To better understand ColeValleyGirl's answer and comments about expanding the selection of named-period tags, I wanted to think more specifically about what this would entail. Such named-period tags would typically be tied specifically to a certain geography.

For the UK – that is England, Wales, Scotland, Ireland (pre-1921) and Northern Ireland (post-1921), the named periods on a genealogical timescale might be:

Now this seems really nice, because it so happens that many of these periods coincide with events of genealogical significance, such as the introduction of civil registration in 1837. The ties between family history and the history of nations can be easy to overlook. Genealogists will also often have expertise or interest in one or more of these periods, which enables them to filter down to just that period rather than just a broad century.

Obviously these periods would only apply to the UK. I started trying to form a list for the United States and Germany (our other two most commonly asked-about countries), but really struggled as there don't appear to be widely accepted named periods. If anyone is willing and able to make an example of named periods from these or other countries, please add them here.

My biggest fear is that we would be constantly checking this list to see how to properly re-tag questions. If the century tags also still exist, then this raises the question of how will users know which to tag?

6
  • This is more granular than I had envisaged, and I'm not the value of the granularity is worth the effort it would involve in 'eduation' and 'policing'. I need to think... – user104 Mar 26 '17 at 17:40
  • Ok that's fine, I just wanted to put this out there so we can get a better idea of how to break down named periods if that's the course we're going to take. Feel free to edit this post to make it less granular. I'll edit out the sub-periods (Jacobean, Regency, etc) for now. – Harry V. Mar 26 '17 at 17:53
  • I see the switch over from ruler based periods to war based periods at 1914 to be at least as arbitrary as my proposed switch from decade to century tags being preferred from 1800 back. I have no objection to a few well-known periods (i.e. those associated with major conflicts) being given priority when time tagging but I think decade/century tags get a big boost when we think about our global user base. The edwardian-period tag seems less likely to be used by Australians than a federation-period tag for a similar time period. – PolyGeo Mar 27 '17 at 1:54
  • 1
    Having reflected on this overnight, I definitely think the added complexity it would introduce (especially when extended to multiple geographies each with their own set of 'eras') outweighs any increased utility. Centuries seem a workable compromise, especially as we'll need to retain them anyway for those geographies we haven't got round to yet. – user104 Mar 27 '17 at 7:07
  • I do take issue with @PolyGeo that the nomenclature is arbitrary -- it was a standard part of the school curriculum way back when I was at school. – user104 Mar 27 '17 at 7:09
  • 2
    @PolyGeo I agree it is somewhat arbitrary however this list was based on Wikipedia's Periods in English history – Harry V. Mar 27 '17 at 9:46
-1

Rather than being too purist in our approach to tagging questions for the time period of relevance to them, I suggest a pragmatic guideline of using time period tags in this order of priority:

  1. Named time periods when they are more or less recognized universally (at least in the English-speaking world) e.g. , ,
  2. Century when 18th century or earlier
  3. Decade when 19th century or later because by the 19th century governments begin to take a census every 10 years suggesting that too much changes in 10 years to assume that what was true 10 years earlier still holds.

Users will not need to know that decade tags prior to the 19th century should not be used if each decade tag in the 18th century and earlier is made a synonym of its century tag.

10
  • 1
    Why have you arbitrarily decided that decades only are appropriate in the 19th century and beyond? How will users know which type to tag without reading this post? Maybe you could include example usage guidances for the three categories that makes this clearer. – Harry V. Mar 26 '17 at 11:08
  • @HarryVervet There seemed to be resistance to the idea of decade tags in the 17th century, and the pace of change in the 18th century seems to have been only a bit faster. By the 19th century governments begin to take a census every 10 years suggesting that too much changes in 10 years to assume that what was true at the previous census still holds. – PolyGeo Mar 26 '17 at 11:25
  • Users will not need to know that decade tags prior to the 19th century should not be used if each decade tag in the 18th century and earlier is made a synonym of its century tag. – PolyGeo Mar 26 '17 at 11:27
  • I think we risk chasing our tails on these time period tags, and that the way we have been using them previously works well enough, even if that has been less than prescriptively. In this answer lies a simple formula for deciding what time period tags to use. – PolyGeo Mar 26 '17 at 11:34
  • 2
    As @HarryVervet says, this isn't a simple solution, as it either requires people to understand this discussion, or it allows people to add meaningless tags that will be replaced by synonyms (which won't encourage their understanding either). Further, just because individuals' locations and family structures may have changed every 10 years since some variable date in the 19th century (as revealed by the censuses) that doesn't mean that the relevant genealogy techniques and records changed. – user104 Mar 26 '17 at 16:09
  • Or would you only use decade tags for places in the periods they had censuses every 10 years? -- which is (to put it bluntly) nonsense. Rather than chasing my tail (as you put it), I and others are trying to reach a decision about a simple, practical system of tagging for time period (even if it isn't 'purist') that is clear to newbies and addresses the limitations of our current systems. If you think the limitations are less important than the ability to filter by decades in some but not all periods, and our current system works well, explain why you think the limitations don't matter. – user104 Mar 26 '17 at 16:10
  • Having read the answers from the community, I now wonder if we should have tags referring to time periods at all. Yes, this is partly a tongue-in-cheek answer, but I keep coming back to the idea that the tag should be able to stand alone as the only tag for the question. In some of the scenarios I was thinking about, the time period is implied by the type of record being asked about (e.g. civil reg in the UK) and thus the time period tag is redundant. – Jan Murphy Mar 26 '17 at 18:47
  • @ColeValleyGirl "Or would you only use decade tags for places in the periods they had censuses every 10 years?" - of course not, that would lead to a ridiculously complex set of time tagging guidelines, especially compared to what I have proposed. – PolyGeo Mar 26 '17 at 21:12
  • @JanMurphy I think time tagging is valuable, it just needs to be practical i.e. using a few well known and more or less global time periods, using decades (recently) and centuries (further into the past) to "fill in the gaps". I think new users will understand those few rules, and even when they do not time tag "correctly", it's very easy to re-tag as long as the guidelines are kept simple. – PolyGeo Mar 26 '17 at 21:18
  • 1
    @JanMurphy There must be something in the water! I spent some time last night debating with myself whether we needed time tagging at all, which confirmed my earlier conclusion that we don't when we're taking about some specific datasets.But if we don't know what datasets we're talking about, then we may do (e.g. 19th-century so that somebody who is trying to understand what their options are for locating birth information in the 19th century are can filter down to questions that may be related. – user104 Mar 27 '17 at 7:16

You must log in to answer this question.

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