I don't know how to do it in iD, but the wiki lists some ways of doing it in other editors:
osm.wiki/Change_rollback

Also, there's lot of other useful advice on the wiki. Some pages that might be particularly relevant to you are:
osm.wiki/Using_Imagery
and
osm.wiki/Roof_modelling

To conclude, this changeset seems to worsen the mapping of these buildings, and I think it should be reverted, unless you have any good reason that it shouldn't be reverted.

Hi daniel,
I've been looking at a few of your changesets (this is something I usually do when commenting on someone's changeset), and I think there is almost certainly a problem with this changeset too.
Before this changeset, the buildings mostly had perpendicular walls, which appears to be correct based on aerial imagery here (and the ones that weren't perpendicular on the map possibly should have been). Your changeset has now made the buildings a lot wonkier, with many walls no longer being perpendicular. In general, if a building has perpendicular walls in reality, then you should ensure this is reflected in the map, and I think most editors have tools to help you do this accurately.

Additionally, the general effect of this changeset has been to move all the buildings a small (though inconsistent) distance to the south. This sort of change often means that either you or a previous mapper have used aerial imagery that is not correctly aligned. The original mapping (before your changeset) seems to be correct relative to most of the surrounding buildings, and the public GPS traces don't clearly show that the original mapping was misaligned, so in that situation I think it is best to keep the realtive alignment of features correct, rather than trying to move a few buildings to match individual buildings to match aerial imagery that could well be misaligned.

Hello Daniel, and (a slightly belated) welcome to OpenStreetMap

I'm a little confused by this changeset. Firstly, this is because your changeset comment doesn't say anything useful, and certainly doesn't help me or anyone else to understand what you're doing. You also haven't given any indication of what sources you've used to take make the changes - there's a big difference between how much other contributors can trust a changeset if it's based upon someone visiting the location, compared to if it's based upon out-of-date or hard-to-read arial imagery. Another thing that would also help is splitting this up into two separate changesets, since the changes in London are unrelated to the changes in Hull.

Secondly, the changes themselves seem a bit odd. The change in London appears to be a realignment of the outline of a building under the station, but this new alignment doesn't seem to follow any building edge that I can deduce from other mapping and arial imagery. You also set the layer to 2 for some reason, even though the building seems to be under the tracks (which I think are level 1).

The changes in Hull also seem surprising. Has New North Bridge House been demolished? I can't see any evidence of this online. You also created new nodes for a building shared with an underground cable, which might have been accidental and is probably wrong - I wouldn't expect an underground cable to be running directly under an external wall.

Can you clarify what is going on with this changeset, and fix any errors you think you've made?
Thanks,
ACarlotti

This look wrong - are you sure your sources are up-to-date? The recent A14C2H works were supposed to close up all accesses like the one you just readded, with access provided from the A1307 (or other parallel routes) instead. In the unlikely event that there is still an entrance here, it certainly wouldn't provide access from the M11, because that would require a sudden left-turn across two lanes of 70mph traffic from the A14.

You seem to have mistakenly added a node to the residential area boundary. This probably shouldn't be sharing any nodes with the track you added.

This just duplicates the tagging that's already on the building outline, and is therefore redundant (c.f. "one feature, one OSM element"). I think you should revert this changeset.

Hello Chrisgf98,

Could you please add meaningful changeset comments to any future edits you make? These give a clear indication to other people (including you in the future) of what you intended to do, which makes it much easier to avoid misunderstandings, and correct any mistakes more quickly. There's more information about this on the wiki at osm.wiki/Good_changeset_comments
(To help people understand the changes youve already made, it might help to comment on your existing changesets with a brief description of what changes you have made, and what sources of data you have used. This latter point can particularly helpful if people know that a source you have used is inaccurate - e.g. if you have mistakenly used out-of-date aeriel imagery.)

I also notice that three of your changesets seem to include edits made in two or more locations that geographically a long way apart. This is generally a bad idea, because it is very hard to work out what areas are affected by changesets when they have a large bounding box. (Some tools only show a bounding box, and some other tools that try to show all the changes break if the bounding box is too big). In future, could you please try to ensure that changes in different locations are uploaded in separate changesets.

Thanks,
ACarlotti

I think the tagging is now better, but there still seem to be some inconsistencies. Firstly, you used the tag "oneway:bicycle=-1"; I think this is an error, since I can't find that tag listed on the wiki anywhere.
Secondly, I think there may still be issues with tagging for pedestrian (and equestrian?) access along the cycle routes, but I don't have the local knowledge to be sure. The easiest way to test this might be to wait a few days for route planner caches to be updated, and then test out planning routes in both directions on both sides of the junction (if relevant); however, this is not a guarantee of correctness, since I'm pretty sure at least on of the route planners is misinterpreting some access tags. Alternatively, jsut inspect the routes to see if the tags permit the necessary pedestrian/cycle access.
The issue with the southbound carriageway was that it previously had a tag of "access=no", which you've now removed.
Anyway, thanks for looking into this issue so quickly.

Could you please review and fix the access restrictions tagged on Princess Road (and possibly some of the surrounding roads). You're changes in this changeset and changeset/56900652 (there may be other relevant changesets I haven't noticed) appear to have incorrectly prohibited access both for cars continuing along Princess Parkway, and for cyclists along the cycle route to the east (in particular along the section shared with buses).
Additionally, could you confirm what source you used to establish the existence of a cycle prohibition?
If you need any help working out a valid set of tags, then I can probably help; however, I have no knowledge of the local area, so can't fix this myself without making guesses that might be wrong. From what I can work out, however, I think the intent of your changeset might be best represented by tagging the through stretch of Princess Parkway with "bicycle=no" (and an implicit "access=yes"), and tagging the bus route with "access=no" and "bus=yes", and "bicycle=yes" where it is also used by bikes. I suggest you check this against the advice in teh wiki though, because it's possible I might have misremembered a bit of this.

Should the individual ways now be tagged as "railway=construction" and "construction=rail"? This would then cause them to appear on the main rendering on the website (similar to how the line from Bletchley towards Bicester is currently shown).

Hello michellefarns, welcome to OpenStreetMap, and thanks for your contribution.

There are a couple of small issues I've noticed in this changeset. Firstly, there are a couple of nodes (node/2085120254 and node/27234954) where you have removed some of the tags, with the remaining tags not making sense by themselves. If the café and the car park no longer exist (and are not just temporarily shut), then these nodes should be deleted. Otherwise, the tags should be changed to include at least one tag specifying what type of thing the node actually represent.

Secondly, this changeset covers (at least) two separate unrelated groups of edits, which are in different countries. This makes it difficult for people to work out what has been affected by the changeset (e.g. the 'History' option on the website only shows the bounding box). In future I would encourage you to split your edits into separate changesets covering smaller geographical regions. In general, if two edits are unrelated (roughly speaking), then they should probably be in different changesets.

Finally, it is recommended that changesets include a tag indicating the source(s) used in the edit (e.g. things like 'local knowledge' or 'Esri imagery') - this makes it easier for other people to understand why someone made a particular change, and whether it might be a mistake.

Anyway, thanks for helping the project, and if you can fix those two nodes I mentioned, that would be great (and don't be afraid to ask for help if needed).

3: Oh, oops. Good investigating - when I fixed two issues with that boundary in May, but somehow missed the one I introduced myself in October (probably because I didn't look at the history of the split apart of the way).
1: I guess I must have oversimplified it in my memory - despite having looked at plans of the layout in the Northstowe planning documents only the day before. I think your subsequent changes are probably an improvement, but this is probably an example of where using a way to approximate an area is difficult.
4: That might help to reflect how it can actually be used in the current layout. I just had another look at the plans, and the ones on the website unhelpfully have a label covering this location. The other ones I can find just show it coming to a dead stop, as if they were expecting someone else to fill in the gap (as is happening near Northstowe).

Nice work - there are a few things here that I now realise I knew about but didn't notice or got wrong. For one thing, I now see that I misidentified a farm access as Scotland Drove, which explains why I couldn't find the features I was mapping in relation to.

A few minor issues I've noticed:
1. The mapping of the cycleways around the roundabout near Oakington suggests a more irregular layout than exists in practice - I think the mapping should be more symmetric and regular, with the current irregularity probably just an artifact of your GPS trace.
2. There are a few places where you've tweaked the cycleway alongside the A1307 to match your GPS trace, whilst not moving the carriageway, leaving an inconsistent spacing between them. The most obvious example is just east of the Swavesey junction. It would probably be good to make these match up again.
3. There's an administrative boundary running along the line of the A14 near Swavesey that was edited in this changeset. That was probably due to it accidentally sharing a node with a way. This ought to be reverted by removing any such nodes (you might need to check older history if the incorrect intersection has been around for a while).
4. I think the current tagging of the end of the cycleway near Girton will give weird routing for inbound cyclists (and possibly for outbound too) because the access and oneway restrictions force them to doubleback. I do think this is a slightly awkward area to map, and that this is a relection of this currently being an awkward area to actually cycle - hopefully we can get some better joined-up infrastructure here in the near future.
I might have a look at fixing some of these points myself later, but I don't have the time to do so right now.

https://www.cambridgeshire.gov.uk/residents/travel-roads-and-parking/roads-and-pathways/heavy-or-abnormal-loads-on-the-highway has details of all restrictions in Cambridgeshire, though the maps on there probably can't be used directly as a source. (They can, I think, be used as a basis for identifying which locations you'd need to survey.) If there's a definitive description of the restrictions anywhere that's not based directly upon Ordnance Survey mapping, that might also be usable.

For reference, this issue has been reported previously on many occasions, including in
https://github.com/gravitystorm/openstreetmap-carto/issues/2489

layer=0 is implied if no layer tag is present, so an explicit tag is not needed, and doesn't solve the actual problem, which is probably a bug in the renderer.

I'm puzzled. This changeset seems to suggest that the bidirectional cycleway that used to stop at the A505 end of the slip road has been extended under the bridge. However, the recently published list of "COVID-19 Temporary Cycling Proposals" from the Cambridgeshire County Council includes the item "Make [A505 slip road to Granta Park] one-way for with segregated bi-directional cycle lane". This implies that there isn't currently a cycle lane alongside the slip road. Furthermore, imagery suggests that there isn't currently room under the bridge for a segregated cycle lane and twoway motor traffic.

This confusion is made worse by the absence of source and comment tags in all of your changesets (including this one). With a good comment and source tag, I might have been able to work out whether the changeset was correct, or whether you had made a mistake. As it is, I have no idea waht source you used and whether you might have misinterpreted something you saw.

So I have two requests. Firstly, could you confirm the intent of this changeset, and let me (and other OSM users) know what source you used to determine the existence of the cycleway. (Also, if it does exist, then you should check the geometry near the roundabout, as the layout currently depicted would require cyclists to make two very sharp turns to reach the cycleway).

Secondly, can you please add meaningful changeset comments (and source tags) to all future changesets. This makes it much easier for other people (and also you in the future) to understand the changes you make, and fix any mistakes more quickly.

Thanks,
ACarlotti

Oops, I certainly did mean Hartington Grove - I was thinking that but wrote something different. And I meant the crossing on Hills Road, not the one on Cherry Hinton Road.
The discrepancy I'm talking about is fairly small. Judging by satellite imagery, I think the previous mapping had that crossing about 3m too far north (of the centre). You've deleted the old crossing node and added a new one which I think is about 17m too far north. (Incidentally, I think the edit history would be clearer if you'd moved the existing node, since it's the same physical point object).

What was your reason for moving the crossing near Huntingdon Road? As far as I can tell (based on memory and Esri Clarity imagery) the previously mapped location (south of the current one) was more accurate, and even that might have been slightly too far north.