Suggestions to decrease the open edits backlog


#1

Currently there are 829 pages of open edits and from my experience most of them will go through without a single vote and probably many won’t even be looked at once.
I’m already subscribing to 917 artists (increasing every day) and vote on almost all open edits for subscribed entities to do my part and now and then I go to all open edits and vote on some there.

My suggestions:

  • an additional editor group between normal user and auto-editor (similar to what was suggested here). Their edits would only be open for e.g. 3 days and would only need 2 unanimous votes.
  • different waiting periods and vote-thresholds for different types of edits.
    • e.g. Record merges by @jesus2099’s script basically only need to be looked at once to rule out an error in the script, so these should only need one positive vote.
    • It would be great if edits where you corrected your own error wouldn’t need another 7 days to go through.
  • ability to display open edits by edit type, so you can vote on edits you are familiar with (e.g. only display record merges, only added cover art…).

I think these measures would decrease the list of open edits and more attention could be directed to edits that really need it.
E.g. By chance I just found a series of 24 messy edits by a a new user when all they tried to do is merge two entries of the same artist (legal name and artist name), but obviously hadn’t found the merge function yet.


#2

Reducing the time window for voting would reduce the size of the backlog, but that isn’t a particularly important metric. There’d be the same rate of edits to vote on, but people would be more likely to miss edits that they might have wanted to vote on and IMHO the data quality would suffer. I find the 7-day window about right.

Not always; there are edge cases here, where it’s not immediately obvious.

Agreed. If the database could revert a change, conditional on the reversion having no side effects, that would be a nice feature. But tricky to get right!


#3

People would be more likely to miss edits by rather trustworthy editors and therefore miss less edits by new members (provided they spend the same time going through open edits).

There are always edge cases. And even with 3 unanimous votes you can still make mistakes, but I think it would make sense to concentrate less on cases where mistakes are rather rare (and risk missing a few of them) and more on cases where mistakes are quite common.


#4

For me, I came here from Wikipedia. I know that the two sites serve different purposes so I don’t suggest “why not do what they do”. But it was nice that I could review an entire edit and I could let it be, undo it, or make an individual change (instead of a complete undo).
Here, on MB, everything is an individual edit. When I add personnel to an album, instead of making one edit, it gets counted as 40-100 edits. Let’s face it, if I am adding personnel to an album, we don’t need to review 100 individual edits. It is one edit.

.

My one suggestion for expediting the open edits would be to make search easier (user friendly, aka search for dummies).
When I am on the ISNI or IPI website, I would love to be able to see only open edits that involve those edits so that I can verify while I am there. A quick easy search function could show me all ISNI changes.
Search could show me all merges and/or splits. Then I could vote if John Smith is the same John Smith.
Search could show me all date changes. If John Smith was born in 1812, but someone changes the date to 1776, you know they are talking about a different John Smith. But changing it from 1812 to April 5 1812 could be quickly found and approved.

Also, being able to separate the type of open edits would make keeping up with the time line easier. Because when I am on page one of ALL open edits, by the time I get through looking and voting, those edits are now on page 3. But if I was able to separate the edits, I could go and vote on all ISNI changes that are on page one, and then move to page two because the work flow wouldn’t run at a negative timing.


#5

I really don’t want the vote period being shorter.

I subscribe to my collection and this way I mostly check edits made to entities that I have myself added or checked and I want them only to improve and not degrade.

With 7 days period is already too short for me to get a chance and not miss when I’m in holidays.

IMO the amount of open edits is not a problem as long as people wanting to monitor this or that part of them have enough time to get a chance to do so.

You should rather look at that as a review time frame for all edits than a total list of edits in a queue which aims to be always 0.


#6

I may agree with you on that (the non-vote wait period).
But do you have any suggestions for how to speed up the voting process so that more items get voted on, which would have the effect of shortening the vote period.


#7

I mean even if there are three voters at once right away, I don’t want the edit to pass because, me, I would not have had the time to review the edit on my collection.
Without mentioning fake voters.
I do not want anything that would shorten voting period, at least on destructive edits on my collection/subscription. :sweat_smile:


#8

I personally don’t see a problem with thbere being a lot of open edits.

However it would be very convenient to be able to filter edits that need my attention from those that don’t - auto editors, for instance, I probably don’t need to see any of their non-destructive edits or recording merges etc, whereas I would like to see anything done by a new user.
It is tedious scrolling through everything just in case.


#9

Yes, filtering edits so that those most likely to need attention get the most attention would be good.
Seeing the number of open edits makes it difficult to maintain any sense that voting on some tiny fraction of 1% of them is useful.


#10

I also don’t want shorter times, especially not for deletes or merges. Heck, I’d want those to remain in the queue forever if they don’t get a preponderance of votes either way (even then with a 7-day wait once the threshold is reached; cancel and approve are available as expediting measures).

I do think it makes sense to be easily able to get a list of just those edit types.

And yes, a Revert option (sort of a cancel-after-the-fact) for simple change edits (as long as there is no later edit for the same thing) would be nice, but likely tricky.


#11

It’s clearly impossible for single persons to monitor the full list of all open edits. No matter if it is 800 pages or 400 or 200.

IMO, the better approach would be to try to increase the number of edits which are covered by subscriptions of experienced users. I have subscribed to various releases, artists, labels series etc. which I care about. Thus, I’m inherently interested in edits going on there and hence I will usually check them carefully when I get noted. Also, I have some knowledge on my subscriptions which gives a good basis for evaluating the edits. This doesn’t necessarly mean that I will always vote on them, just out of efficiency considerations.

The current subscription system certainly has some shortcomings. Notifications come separatly for each subscribed entity, and they often overlap. Which means that checking the edits, you may encounter the same edit several times in different lists, which slows you down. It would be great to additionally get a unified list of all the edits which happend in any subscribed entity.

I wonder which percentage of edits is currently covered by at lease a single subscription. To improve the review of edits, we should try to get this number up.


#12

But thanks to collection subscription, I could remove almost all my other subscriptions and kept only the one for my owned records to monitor.


#13

That almost works for me, but it doesn’t extend to edits made to works linked to recordings in my collection. So I have another collection of works. The trouble then is that I don’t really want to see every edit involving any of those works – just changes to the work data itself. I don’t know a good way to do that.


#14

Agreed, collections are a nice thing, but they only work for releases afaik. That is not enough for me. For example, I’ve put some effort into cleaning up certain series of Deutsche Grammphon, so I subscribed to those series. When some release is edited which is in such a series and in the same time in my collection, that edit shows up twice in my notification lists.


#15

Not only – you can have a collection of series, or recordings, or works, for example.

Does it show up twice in https://musicbrainz.org/edit/subscribed ? That’s surprising (and sounds like a bug).


#16

No indeed, it doesn’t show up twice in https://musicbrainz.org/edit/subscribed and that’s what I use now.
I regularly scan this list for open and closed edits.
But I miss a proper search criteria to be able to scan this list with no time limits as this URL is limited to last 2 weeks or something.

MBS-7912 — New Edit search condition : edit is [not] in my subscriptions


#17

I use this page
https://musicbrainz.org/edit/subscribed
as my shortcut to come to MB. Every time I access MB, that is my landing page.


#18

It is limited to 10 days.

(Edit: Fixed post to have enough characters by making your reply a full sentence, rather than adding enough characters by adding irrelevant filler text. —@Freso)


#19

Then this documentation is wrong. (“Collections are lists of releases …”)


#20

No. The problem is that some edits do not show up at https://musicbrainz.org/edit/subscribed at all. Maybe that is a bug?

I get a daily email with the title “Edits for your subscriptions”. There are parts “Changes for your subscribed collections”, “Changes for your subscribed artists” etc. As https://musicbrainz.org/edit/subscribed apparently does not work as it should, I click on all the links following those headings. Every now and then, the subsequent lists of edits overlap and I see one and the same edit several times (in different lists).