I was wondering we do not have a good solution when the release year is "unknown" and we know just the decade or an estimate year. Leaving the year blank is one solution, but we miss the information e.g. in the MR. Often Master Releases have more than 100 releases, with a third of unknown-year-releases.
On the other hand we can find informations on the release which indicates the earliest year possible,
like Labelcode (1976),
logos which were created on a fixed date and wasn't used before e.g. wmg-logo (1974),
listed company names which were used in a known periode (funny to see in labels discography when releases exists which were released before company was foundend)
or labels (colour, writing, Text in RIM ....) which were used from e.g. 1976 to 1979.
In this case it would be great if we would be able to enter the release for example this way
1976 ++
to indicate the release was issued after 1976 (for sorting purposes, more information can be entered in notes)
another good idea is the idea from this http://discogs.programascracks.com/help/forums/topic/225884
submission warnings if
e.g. the LC field is set and year is before 1976 or CD released 1969 or Vinyl from 1662. (don't know the exact years now)
losAntonis
If the decade is known a year like 198X, 199x, or 200? would be a good idea.
Agreed.
losAntonis
Chronological sorting by actual date, not just year?
Now that seems like the most obvious thing ever, still it's not implemented, drives me crazy to see releases in the wrong order even though the full date is entered...
losAntonis
If the decade is known a year like 198X, 199x, or 200? would be a good idea.
i would prefer something like "1987-1991" enabling the submitter to cross decades.
just today i was looking for a specific beatles reissue which is usually a time consuming task. the submission in question had release year 1987 entered and in the notes field it said something like "these pressings were made between 1987 and 1991". now while this may not be entirely correct according to the discogs guidelines it was really helpful info for me.
i've had cases where i just gave up trying to find the right pressing for say a beatles or a led zeppelin record in the database and ended up NOT listing it for sale in the marketplace. i just didn't want to spend an hour or so to go over all the releases with identical cat-no and "unknown" release date searching for tiny differences that are often not even spelled out in the notes field. this i believe is also a main reason for many listings violations in the marketplace.
Where the exact year is not known, there's no benefit to adding anything to the year. The current system of adding approximate ranges in the notes is better and sort at the foot of the data lists instead of occupying what will be quite random and incorrect positions in the discography.
Eviltoastman
The current system of adding approximate ranges in the notes is better
Could you perhaps elaborate in detail how exactly is it supposed to be "better" than having a dedicated range field?
What are the benefits of not having something?
In other words: to me it sounds like saying for instance "polygamy is better than monogamy". Or whatever… ;)
Eviltoastman
instead of occupying what will be quite random and incorrect positions in the discography.
As opposed to the current positions which are… totally random at the bottom, eh?
Coincidentally just noticed today what a mess it is at the bottom of Doors* - L.A. Woman while I was adding yet another undated European vinyl reissue/repress varation of the same, released in the late 1990s and bought by me probably in 1999.
If a Master Release has only a few releases the problem with the unknown releases is not that big, but if you have a Master release like e.g. Led Zeppelin - Untitled it is very hard to browse all the entries till you find the one you have, or you will add another dupe.
On the other hand it is an illusion to think the entered years of all this releases where years are added are correct and accurate.
A lot of them are not, and it drive me nuts if late reissues have the year of the initial release, while some releases are "Hidden" in the unknown area, because it is not completely known if the release is from e.g. 1972 or 1973 (for a 1970 initial release if only these two year are possible).
I don't want adding guessed or approximate data, but I want to be able to add the year which is most obviouly right and mark it as such with the benefit, that the releases are correctly sorted, we could find them faster, instead of opening 20 to 50 unknown releases till I find the correct one.
Eviltoastman
Where the exact year is not known, there's no benefit to adding anything to the year. The current system of adding approximate ranges in the notes is better and sort at the foot of the data lists instead of occupying what will be quite random and incorrect positions in the discography.
+1 This is one proposed change I cannot . Toasty is spot on this time, and his comments are surprisingly not Evil. :D
losAntonis
I don't want adding guessed or approximate data, but I want to be able to add the year which is most obviouly right and mark it as such with the benefit, that the releases are correctly sorted
If you don't know the actual year it can't possibly be sorted into the correct place.
yes it can be. relative to the other releases. so that one day we might have a proper chronology instead of a complete mess like we have now with bigger master releases.
Fauni-Gena
If you don't know the actual year it can't possibly be sorted into the correct place.
You're in the IT business, aren't you. Ever heard the term "fuzzy logic"?
An approximate (fuzzy) "place" is not an "incorrect place".
In the vast majority of instances we're talking about, the fuzziness can be narrowed down to ±1–3 years. That's actually pretty precise on a scale from 1860 up to 2014.
And a lot more helpful than the Big Bag Of Unknown™ which at this very moment counts whopping 413,318 releases.
There are lots of knowledgeable people here who know what to look for. They just need a place (a field) where to put the data in, so that it can be meaningfully parsed and sorted.
In the release notes that data is useless for any parsing.
I think there are some good ideas here, and something like this would be more than welcome.
loukash
There are lots of knowledgeable people here who know what to look for. They just need a place (a field) where to put the data in, so that it can be meaningfully parsed and sorted. In the release notes that data is useless for any parsing.
loukash
Could you perhaps elaborate in detail how exactly is it supposed to be "better" than having a dedicated range field?
I think I addressed this with the rest of the sentence:
Eviltoastman
sort at the foot of the data lists instead of occupying what will be quite random and incorrect positions in the discography.
Fauni-Gena
+1 This is one proposed change I cannot . Toasty is spot on this time, and his comments are surprisingly not Evil. :D
Yes, well sorry for my recent bad mood here, I thought a break would help because I know I've been quite short, but it has not helped. This isn't a "waaah, I hate you guys I'm leaving (please beg me to stay)" speech, because I loathe those, just as everyone does. It;s just that if I can't shake this funk off I will try to stay away permanently. Something about the place just breeds an antipathy in me which is not me.
loukash
As opposed to the current positions which are… totally random at the bottom, eh?
+1,000!!
I also agree in having 196X, or 197X,... at least we should have releases sorted in approximately their correct place, not at the bottom of all releases - which can also give problems as you can't see your release is there, and you add it duplicated, what I've seen done more than once!
Fauni-Gena
If you don't know the actual year it can't possibly be sorted into the correct place.
If someone don't know the year it should stay empty, but if you know that a release is either from e.g. 1973 or 1974, most obviously 1973, but not complete sure, it is really sad to let this release year unknown and hidden in the unknown section, just because the missing accuracy of some months.
Eviltoastman
otherwise they occupy an arbitrary position within the discography sort.
even with an approximate release date they would appear at the chronologically correct place in the discography. for example if a release has one UK pressing from the 60s, one from the 70s and one from the 80s then sorting them accordingly is quite the contrary of arbitrary.
From my background in archives, I used to use conventions for expressing unknown or conjectured dates, such as square brackets [1964], question marks [196?] and approximations [c. 1964]; similar methods still seem to obtain in the world of digital library catalogues, if this page http://oclc.org/bibformats/en/2xx/260.html is anything to go by, which suggests there may be some way of normalising such fields in a database.
My suggestion is:
~1973 = approximately 1973, the span has to be declared at the release notes
~197? = a 70s release
The primary syntax should be ~ as it would be understood by everyone without clarification.
It should be mandatory to state clear notes for the reasons. At least we need some labels be tagged at the profile to state clear notes for the label design, as there are usually several version in some periodes, for example RCA Victor and similar.
Think the questionmark at the end for the decade has been repeated so often here, that is clear, we want it.
The addition here is off course the primary syntax ~ to both, the approximately year, as well as the decade. A confusion is not possible than anymore.
Eviltoastman
sorting at the bottom is much better, otherwise they occupy an arbitrary position within the discography sort.
Even after a "Fuzzy Date Field" is introduced, you can still tell the developers to create a special "Eviltoastmansortingoption": [√] Ignore Fuzzy Date when Eviltoastman sorts by year
Nothing easier than that. :P
+1 Very good ideas in the first post of this thread.
All asked for before, as he showed as well.
Eviltoastman
The notes are fine, sorting at the bottom is much better, otherwise they occupy an arbitrary position within the discography sort.
I really don't see the benefit of chaos above "approximately correct".
Having this information in the notes is fine too but useless for search and sorting options, I thought you would see the benefit of that as well :).
hermanito
I really don't see the benefit of chaos above "approximately correct".
Having this information in the notes is fine too but useless for search and sorting options, I thought you would see the benefit of that as well :).
My thoughts exactly.
It's often impossible to pin down an exact year for represses and reissues, especially when done by the original label. But there are plenty of clues which help us pin down a range (changes in company details, pressed at a particular plant which was only active for a certain date range, addition of SID codes, etc.)
I've got a lot of releases where I can't be certain of the exact year, but I can be certain they were not before and/or after a given date. I think this is useful information and very much worthy of being entered in a searchable manner.
manus-von-alles
My suggestion is:
~1973 = approximately 1973, the span has to be declared at the release notes
~197? = a 70s release
problem is, certain label designs crossed the decades, like the european red cbs label with white "cbs" on top that was in use from 1985-1991. "198?" may not be correct in this case so something like "1985-1991" should be allowed too.
Especially for a CD whose matrix confirms it to have been pressed, eg, in 2013 but was not submitted until April 2014, the option to add and sort an "approximate date" of 2013-14 would obviously be a lot better and more helpful in MR view than having it mixed up with other dateless releases at the bottom. So I give +1 to all those in favour of having this range of dates option.
Good idea.
I am adding lots of 60's singles lately and sometimes it's impossible for me to find an exact release year.
Things like "1963 or 1964" happen often and it would be nice to add this info, instead of leaving the field blank.
It is lost information and lost work/time.
"1963 or 1964" to notes is no solution.
Yes, but it's easier to use - or / that aren't Alt+Gr based, I think...and most people don't know how to use it; and to copy and draft you should have first to find a "~" (copied form your text now)
;-)
Perhaps technically this would be easiest to implement with having an additional column for the end date of the range and if the additional column is empty, the date is exact... But it can be implemented on the UI side in many different ways – it could even multiple ways if desired. Though, perhaps just adding another input field for the end date with explanation that if used it makes the date to be a range which should be used when the exact release date is unknown.... would be the easiest to implement and use too.
Eviltoastman
The notes are fine, sorting at the bottom is much better, otherwise they occupy an arbitrary position within the discography sort.
If you don't know the exact year, it may as well be e.g. 196? or left blank. I.e. instead of a range.
In this way, all releases with exact year would come first for each decade, leaving the somewhat uncertain last for each decade, and totally uncertain last in the whole MR.
Additional knowledge about range could be put in notes.
Calle_jr
If you don't know the exact year, it may as well be e.g. 196? or left blank. I.e. instead of a range.
In this way, all releases with exact year would come first for each decade
problem is, release schedules didn't always factor in the boundaries of decades. what about a release that we know came out between e.g. 1968 and 1972?
i prefere the date range solution because it allows more precise information.
example: currently i'm going through a bunch of german elvis 45s. for the first pressings exact dates are ususally known. reissues can be dated roughly within a range of 2 or 3 years following the label variations. a date range is not only much more accurate than just naming the decade but it would also allow to list different pressings of a release in chronological order.
djmushroom
what about a release that we know came out between e.g. 1968 and 1972?
It would be submitted with year "196?" and the note "release year between 1968-1972 based on center label design" would be added.
For those who prefer, either blank or "19??" could be added plus above note.
I like the error checking options - like not allowing a year before 1994 if you have IFPI codes, or getting an error if you try and add them to a release already listed before that date: "Please check with original submitter if the release entered has IFPI codes" or something.
Listing ranges or possible ranges sound like asking for trouble, the Release and/or Submission notes are fine enough for that kind of nebulous Fuzzy Data.
We already have a certain fuzzyness of release dates. A known year or even a known month are only approximations of the exact release date anyhow. Most of them based on dates that the releases themselves carry for other purposes than reporting exact release dates. Thus I don't think that adding the capability to estimate release dates with larger uncertainties will be detrimental. Sorting the results will be a problem though. Which is likely the limiting factor for
A range would be as easy to parse for sorting if we're thinking about the simplest possible sorting method: The sorting could just use the start year (which could even be the same normal field if the end year had the extra field).
The difference is the range has more exact start year than the "196?-system".
Ok, whatever looks best.
The range will result that the roughest ranges will come before the narrower (more exact) in the MR list, e.g. 1970-1979 compared to 1972.
But I would welcome any of these two improvements.
The range method is very solid when you have a reference to the dates for center label design.
The ?-mark method is more convenient when your reference is record catalogs which often use like "196?" or "70s reissue".
However, all of this would require a solid release note for the reasons, and this is vacant in near all cases, so I don't see any chance or reason to introduce this guesswork. Cheers
manus-von-alles
would require a solid release note for the reasons
Well, noone hides that the reason for this improvement would be to allow for somewhat more approximate dates. That it's better to know something than nothing. Dates wouldn't be wrong just because they are approximate.
The decisive question; is it better to have a mix of exact and approximate dates, or no dates at all?
Calle_jr
The decisive question; is it better to have a mix of exact and approximate dates, or no dates at all?
I don't see any problem to denote the approximately date at the release notes. So we can hardly talk about no information at all.
To state an approximately date you needs solid knowledge in label designs, company periods and some stuff more. Most are hardly able to distinguish between the release date and the copyright. Let me tell you, I fixed thousands of incorrect dates for this reason.
So a first step would be to update the release notes with proper information for the approximately date. Maybe we once find a possibility to improve the system?
Agreed, I only think this is a good idea if we use the same label layouts chronology, so that should be documented here properly first. There are actually quite some websites already documenting this with big labels.
I do really like this idea though, sometimes half the subs in a master release don't have a release date.
manus-von-alles
Most are hardly able to distinguish between the release date and the copyright. Let me tell you, I fixed thousands of incorrect dates for this reason.
I know what you mean. Reissues/represses sometimes have the original copyright statements with the original years.
But you didn't mean we need external proof for all releases, did you? Of course the release year is usually the same as in the copyright statements when the release is an original instead of a licensed, reissue or repress.
If every release year must be proven beyond any doubt whatsoever with official references it will hurt the database because people start to leave out correct release years...
I don't think you meant that but just in case. :)
Edit: One promo had 2002 in the copyright statements but release year incorrectly as 2003 because the retail version topped in the chart in Jan 2003. A strict policy for external checking wouldn't necessarily improve the correctness of the database... :)
Calle_jr
I think it's psychology. Currently, Discogs are trained to not put any dates because the date is wrong.
Member who are not able to state any date are also not able to state approximately dates. That's exactly the problem.
jval
Edit: One promo had 2002 in the copyright statements but release year incorrectly as 2003 because...
Sorry, what you are talking about? A who is not able to distinguish between a date when a release topped the charts and the release date will generally have problems to denote proper dates.
RSG §8.1.2. Usually, the latest copyright date that appears on the release is the one to use. If there is any doubt, consult other sources, and use common sense. Re-releases of older material sometimes do not have updated copyright dates on the packaging, but you should do your best to find the actual date of that release, not of the original recording.
This is a short information, and the problem is frequently, that we don't know whether the release is an initial one or a reissue, cause label images or other necessary, and for this reason MANDATORY info > see RSG §1.3.1.a. is vacant. This can for example frequently found on problem labels like Capitol, MCA or Verve with frequently different label design while remain the same cat#.
At least, if there is no C or P date on a release, a source is obviously mandatory due to RSG §1.1.2.
The first step should always be to contribute as much info as possible, what is more important than a guessed release date. Cheers
manus-von-alles
Member who are not able to state any date are also not able to state approximately dates. That's exactly the problem.
Runout etchings/stamps + center label design will in really many cases narrow the possible release years down to 3-4 years, while it may be close to impossible to tell the exact year.
Since you are a serious collector, I am honestly surprised that you don't agree to this.
I agree that the full data reduce the possible release dates commonly to a bunch of years. There is no problem to denote this at the release notes. But have a look, you hardly ever find proper info about this there. Cheers
I guess you were trying to think how it was related to this subject. But it wasn't. I was just saying that in the general sense. Off topic, and thus confusing, sorry. :)
On topic, I think you're saying a range without the proof provided would cause problems because we wouldn't know whether it's correct or not. Well, unproven information can always be removed... So that wouldn't necessarily be a problem.
If your point is the proof would be so rarely provided the feature wouldn't be that necessary that can be true.
Edit: Though, not all cases require external investigating. The submitter often knows when the release was bought.
There's one possible problem with the range. A theoretical example: Someone bought a re-release in 1998 and marks the release year range as 1995 – 1998 (saying it was bought in 1998). Then someone else comes and changes that to 1996 – 1998 without providing any proof or explanation (ok, explanation is of course obligated by the guidelines but that will of course happen anyway sometimes, and once the information is there it's a matter of correctness instead of how much the edit followed the rules, you know). Now, how to judge which one is more correct; 1995 or 1996... :) Ok, if there's no knowledge, then 1995 is naturally better. But think it this way: Someone can change it to 1994, then to 1993... There can be kind of "edit wars" when the correct start year can't be determined precisely... In that sense a more rough 199? type of solution might be kind of better. It's just that if it's about, let's say 1988 – 1996 then that won't work very well... A ~1990 type of solution might work for those cases if we're fine with the idea that there can be years which are only estimates instead of precise. Ok, even then people can disagree about which is the best estimate but then again, it's not necessarily a problem because naturally if the estimate is ~1994 and someone comes and says release was bought in 1993 then ~1993 is clearly more correct as the estimate than ~1994. Though, sometimes people would argue about things like ~1990 vs. ~1991 etc. You know, when no solid proof exists.
A solution could be simply 199? and 19??. If a release is roughly from 1988 – 1996 you could then only set the release year as 19??. Before someone argues it's then useless: It isn't. It's still better to have at least the correct century then no date at all. E.g. there can be releases from 1980s, 1980s – 1990s, 1990s, 2000s and 2010s. Then naturally values like 198?, 19??, 199?, 200? and 201? will be useful. Though, those 19?? wouldn't sort correctly. Because of that it might be a good idea to allow them to be entered as 199? even though they can be from the end 1980s... Of course the best way would be to say "end 1980s / early 1990s" somehow numerically... This is a difficult feature to implement properly (almost impossible because the UI wouldn't probably be clear to use)... :)
–> I think the only solution which would technically work is the range. It might be ok if there are clear rules how the range should be defined correctly. You know, it should be as wide as possible. Then there wouldn't be those "edit wars". E.g. if an original release was released in 1991 and you have a re-release you bought in 1996 you should then simply set the release year range as 1992 – 1996. It's then set to a value which is logical and correct. Or in case of an external source like e.g. a record catalogue: "late 80s / early 90s" –> 1985 – 1995, "90s" –> 1990 – 2000, "early 90s" –> 1990 – 1995, "mid 90s" –> 1992 – 1998, "late 90s" –> 1995 – 2000. So, I still think the range would be the best solution if this is implemented. It just requires good rules.
I think both the ?-mark method and the range method would work from a technical point.
The basis should come from: How do we find the date?
- Release ad.
- Music review.
- Date on record (runout) / inner / sleeve.
- Record catalogs.
- Label design catalogs.
- Date of purchase.
Each of these alone are more or less uncertain in case of RPs and RIs.
But when combined, they make rather solid proof.
soundprojector
From my background in archives, I used to use conventions for expressing unknown or conjectured dates, such as square brackets [1964], question marks [196?] and approximations [c. 1964]; similar methods still seem to obtain in the world of digital library catalogues, if this page http://oclc.org/bibformats/en/2xx/260.html is anything to go by, which suggests there may be some way of normalising such fields in a database.
I've once before suggested that discogs should hire an archive professional. It's not like these questions have never been dealt with before. Archivists have been handling these kind of issues for decades. Why not learn from those who know?
Approximate dates and date ranges are much better than "unknown" which is often factually untrue.
losAntonis
I was wondering we do not have a good solution when the release year is "unknown" and we know just the decade or an estimate year.
I agree with you, more than often we don't know the exact release date of an item but we can estimate it from ancillary data such as companies names, matrices styles and IFPI codes. One easy way to handle this would be to store unknown dates as a range: not before this date and not after that other date. This way we could narrow the probable release moment of an item without the use of a complex syntax.
