OpenStreetMap
Recently it came to my knowledge that it’s preferably to upload almost every change you do, instead of doing several changes and then uploading them, because this raises your amount of changesets, which is what other are looking into others. They don’t look how many changes you do in OpenStreetMap, but rather how many changesets you have.
What I was doing till 2 days ago was pretty much 1 changeset = ~80-100 changes, mostly because the iD was going red after 100 changes. So, even though I’ve done a lot of changes, for example in my area, to another person it may not look enough to be considered.
Now I do 1 changeset = ~10 changes.
The reason I was doing that previously is because of my way of thinking for Wikimedia-related edits. Few and massive edits rather than several and changes of like a sentence at a time.
Discussion
Comment from highflyer74 on 9 September 2020 at 14:15
Hello!
I’d say (from a perspective of someone doing QA from time to time) the size of the CS is not so relevant.
I rather prefer CSs that stay local and don’t spread across half the globe. Also I would consider putting different topics into different CSs, e.g. addresses, POIs, buildings etc. And of course good CS comments help a lot to understand the mappers intention.
And I would rather use JOSM instead of iD ;-)
Comment from Sanderd17 on 10 September 2020 at 07:34
There are more insightful stats than just the number of changesets. Many people like to take a look at HDYC to check how a contributor is doing: https://hdyc.neis-one.org/?jimkats (nice history btw, a very active start).
You should try to * Limit it in time: at max a few hours, to avoid conflicts * Limit the area: don’t make continental or even country sized changesets, it will cause many people to see it in their history and investigate it * Just edit one topic per changeset: if your tagging is wrong, and a changeset should be reverted, you don’t lose the other changes.
There’s no hard limit of 10 or 100 changes. I even have changesets of over 1000 changes. These large changesets can happen when tracing rivers or streets in unmapped areas. They’re still just one feature and a limited area, so it should be fine.
Though every new mapper can make the mistake of letting a changeset become too big. I made that mistake too.
Comment from benoitdd on 13 September 2020 at 14:51
On iD, i had a few bad experiences with big sets (>50) being lost after a refresh of the page.
So I’d recommend smaller sets to limit the risk of losing your work and track of what you were working on.
I find it also easier to correct errors on smaller sets when you submit them.
Comment from skquinn on 13 September 2020 at 22:35
It depends on what you are doing. If the entire intent of my changeset is just to add a few traffic controls, fix alignment, or add the address of something (say, a c-store) I’m content with <10 edits. If I’m adding buildings with MapWithAI, 1000+ edits is not uncommon and I’ve even maxed out a changeset at 10,000 changes before.
Comment from skquinn on 14 September 2020 at 08:11
Re: iD losing changes – or, as an alternative, use JOSM for larger changes and iD only for small changes and turn restrictions (which is what I do).
Comment from Foomandoonian on 16 September 2020 at 17:48
Thanks for your comment on my diary yesterday!
This is an interesting question. I think my instinct would have been to save an new changeset every time I finished a discreet task, whatever that may be. A dozen edits on one street, or just one quick amendment elsewhere.
Comment from jimkats on 16 September 2020 at 19:17
This makes sense to be honest. One changeset for a specific type of edits, so to be easier to track down in the future what edits and where those are. Or even multiple types of edits, but within a specific area, in a style of segmenting a bigger area.
I like your thinking everyone, it’s amazing to read such logical ways of editing and at the same time to view several random changesets viewable in my area because other people like to do edits in different part of the Earth in the same changeset. :D
Comment from CjMalone on 19 September 2020 at 10:54
Big changesets (as in number or nodes added/edited) make no difference, If you are tracing ten buildings or a hundred per changeset it doesn’t matter as long as the tooling supports it.
Bigger as in area is more controversial.
I’ve been “told off” for doing bigger (as in area) edits before, so for a long time I’ve been doing a changeset per source, per area.
Now I’m starting to doing bigger areas as long as it makes sense, it’s easier to understand than tons of small changesets, and easier to revert if I make a mistake or do something even slightly controversial.
That being said, I wouldn’t add buildings around a country in one changeset, but I would fix a tag.
A changeset, like a git commit, should tell a story.