From MusicBrainz Wiki
< User:Hawke
Revision as of 23:02, 5 January 2014 by Chirlu (talk | contribs) (→‎Additional comments: Add link to similar idea)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

“Circa” dates

reference MBS-2954

Currently there is no way to store an approximate date in Musicbrainz. This applies to performance dates as well as artist birth/death dates.

The ticket above lists a couple of options for how to deal with them.


Option 1: Circa checkbox

A single “circa” (or “approximate”) checkbox for each date. This would indicate that listed date is approximate:


In AR displays it would be:

  • composer: Wolfgang Amadeus Mozart (classical composer) (circa 1787)

Option 2: six checkboxes

Six checkboxes: “circa”, “before”, “after” for each of start/end. This would give finer-grained details on any start or end date.


Option 3: dropdowns

Similar to option two, but with mutually exclusive options instead of checkboxes:

circa - option 3.png

Votes and reasons

I prefer a single “circa” box because…

  • I just want a way to say 'the dates we have are approximate'. “circa” does this, without needing to worry about whether a given date is the earliest or latest possible. This is also in line with standard use of circaHawke (talk)
  • There are issues that the after/before fields still don’t solve, e.g.:
    1. One date with before/after is not enough when you have a range, which is often the case. E.g. in the example of Kurt Cobain cited later on this page, there is a “last seen alive” date and a “found dead” date, and you could only record one of them.
    2. There are even cases where multiple ranges would be required. We sometimes know day and month of a birthdate with certainty, but only an approximate year, for instance.
    3. “Before” for begin dates is rarely useful. We could enter “was born before editing-date” on every single artist and always be right, but it is not more information than leaving the field blank. The same applies for “after” on end dates.
    Thus adding before/after options would not bring much gain over “circa” alone, but be difficult when trying to interpret the data, e.g. when sorting (born “after 1950”, should that go before or after “1952-12-01”?). Replacing every date with a range or a set of ranges might bring more gain, but also much more pain database-wise and UI-wise.
    I’m therefore in favour of a circa checkbox as a “warning flag”, but additional information in as much detail as is available should be placed in the annotation. --Chirlu (talk) 22:49, 5 January 2014 (UTC)

I prefer a lot of checkboxes because…

  • There may be instances where we know an event could not have occurred before or after a certain date. For example, we don't know exactly when Kurt Cobain died, only that he was discovered dead and estimated to have died a day or two before his body was discovered. -- HibiscusKazeneko (talk) 18:16, 5 March 2013 (UTC)

I prefer a dropdown because…

  • I see it as: (circa) == (relatively closely before or after). Therefore, I'd prefer the second, but with a maximum of one of the three (really, four, adding exact) options available at once. This makes the AR display more straightforward as well: ({before |after |circa |<nothing>}{year}{-month}{-day} to <same>), e.g. (before 1996) (circa 1284-02) (after 1500 to before 1600), etc. If we get real clever we can change it to "between 1500 and 1600" in the latter case :) A <select> dropdown would seem to be the most straightforward implementation of this. Barring that, I'd at least like to see circa checkbox on both dates; often birth dates are approximate, but after someone is well-known their death date is known exactly. Ianmcorvidae (talk) 19:04, 5 March 2013 (UTC)
  • See above. :) --Freso (talk) 18:46, 5 March 2013 (UTC)
  • If we're going to add an "advanced" feature like this, there's no reason not to go for the one that allows us to get the most precise possible data. Also, sometimes "circa" isn't something we know at all, say, we can know a composer died "after 1553" because that's the last year we have stuff written by him or whatever, but that certainly doesn't translate to him dying "circa 1553". Or if something is released in 1975, it is likely that it was recorded reasonably close to that date, but the only think we can know without guessing is it was recorded before 1975 --Reosarevok (talk) 18:25, 5 March 2013 (UTC)
    • Moving this to dropdown since it seems more reasonable than the checkboxes, although I do think "exactly" and "blank" shouldn't be the same and we should offer both options with "blank" as default (sometimes -often!- you just don't know how precise a date is, but using "circa" for those would be strange).--Reosarevok (talk) 01:47, 6 March 2013 (UTC)

I have a better idea:

Additional comments

I’m fairly certain that most people, given a vague date will put it in anyway because some information is better than none. I know I do. —Hawke (talk)

I think six checkboxes is the wrong approach because before and after are mutually exclusive for a given date (you can't start both before *and* after the same date). If we do decide to support more than just a single circa checkbox, I would do it as a dropdown with four options (something meaning exactly, "circa", "before" and "after"). --Nikki (talk) 19:42, 5 March 2013 (UTC)

Same as Nikki and also I think circa sounds pedantic, I prefer around or about. — Jesus2099 (talk) 15:33, 7 March 2013 (UTC) (who wrote circa in that jira ticket)

  • The terms after and before (if it is decided to implement such options) should be replaced with something that includes the given date. To return to the Kurt Cobain example, his body was discovered on 1994-04-08. Few people would like to enter this correctly as “died before 1994-04-09”.
  • In addition to a “circa” checkbox, a choice between different levels of uncertainty might be useful: “±days”, “±months”, “±years” or similar. UI-wise, this could be a drop-down that is hidden until a user checks “circa”.

--Chirlu (talk) 22:49, 5 January 2014 (UTC)

(I now see that my level-of-uncertainty proposal has been made earlier by pabouk as a comment on MBS-1385.) --Chirlu (talk) 23:02, 5 January 2014 (UTC)