Hi Andy,

I hear what you're saying. I've had good experiences with the tool, but there does seem to be a few small UK issues. We may need to look into them further. I think it's probably just a few parameters that need tweaking. I don't think it meets the threshold for automated editing, it's more of an assisted editing. I'm quite surprised at how many issues it's been flagging up. It'd be good to get it tweaked and help clear the massive backlog of work.
Happy mapping!
NG.

Hi Andy,

It was fixes to a lot of random outdated tags in RapId and road realignment mostly. Just small easy fixes to clear out the queue to get to the important stuff later. Nothing of note, just following the Pareto Principle.

No need to apologise. It's great to have someone else contributing. Keep up the good work!

I'm just trying to tidy a few things and make it a bit easier to maintain in the future.

Making as many updates as you do, occasionally someone will come along and edit a couple.

I figure I probably make (at least) one error per thousand edits. Which is a pretty good rate for free labour.

I figure if I get something wrong. Hopefully someone might spot it and tweak it.

It's the ultimate get-out-of-jail free card! ;-)

Hi Catgirlseraid,

Nice to virtually meet you.

The crossings I've been cleaning up, are (almost always) where driveways cross "sidewalks."
Where the right of way clearly favours pedestrians, and being tagged as a pedestrian crossing tag adds a maintenance burden, but no useful extra information for users.

A pedestrian crossing is used to bridge a roadway between two adjacent sidewalks.
Where a sidewalk terminates, and a pedestrian crossing crosses a road/street to join another sidewalk.
Where the primary users of the roadway, are vehicles that have the right of way.
Or to show a safe place to cross the road that may have added safety features, such as signage or crossing signals.

When using navigation in apps, (depending on your config) approaching a crossing may trigger an audible alert to warn the user. This is a desirable behaviour.
Excessive alerts are undesirable, and users will quickly find the excessive alerts annoying, and disable the feature. (When everything is urgent. Nothing is urgent!)
To take the extreme view, there are 1.9 million residential buildings in NZ. If each of their driveways were mapped and marked as pedestrian crossings, there would be an extra 1.9 million crossings/alerts to maintain.

As every driver (should!) know, If they're ever driving over a curb/sidewalk, they have to give way to pedestrians. There's no need to map some driveways/sidewalks as crossings.
I believe an earlier version of some editing software defaulted to this, but has now been corrected in later versions to just join the two features together.

Some of the driveways should probably have been modelled as service roads, and I have changed a few of these where I believe that's a better fit.

Does that make sense?

All good mate. I wasn't worried. Just stumbled across it. Good work sorting it out. Night!

Hi Mate. I've just noticed a heap of points with no tags that seem to be part of this changeset. It's only 3 years old. Any ideas? ;-)

Crossings? Interesting. I've been taking a bit of an interest in them as you may have seen. Just trying to tidy them up a bit, but there's so bloody many! I went through and removed a whole heap that various people had added to driveways. I mean, if you need an alert coming out of your garage not to run people over. You probably shouldn't be driving! Definitely a lot more needs tidying, but I think I need a macro or something I can monitor, as it's hard on the hands doing them over and over. Even just better defaults would be a start.

Hi K. Long time no hear. Cheers for the nice comments. Stumbled across the problem while trying to give the rural some love. I've been having a few Rapid issues since xmas that I've been working with Bryan on. Seems to be getting quite buggy. Might have to treat myself to new hardware. I'd assumed this issue was just old imagery. Don't know why I didn't check the history. Hopefully I'm pushing things in the right direction. Tried the Discord that you suggested. Could never get the captcha to work.

All good mate. No warning at all. I was halfway through adding a few fire hydrants, searched a local street a few kilometres away, and ended up at an identical street in the UK.

I backed out and headed back to NZ, but must have likely moved something by accident on the way.

No warning at all. It probably incremented the number of edits, but as I was halfway through, wouldn't have noticed.

I've analysed it, but can't really think of any way to prevent it happening again. I consider myself very careful, but I missed this one.

Usually I'd get back to NZ by hitting the GPS location icon, but I haven't been able to get it to work in this browser. I might have another look at that.

Or, reverse my last edit, that'd take me home.

Luckily, my error was swiftly noticed and rectified. The system works!

Au revoir! ;-)

Whoops. Sorry about that. I was looking for Prestons Road, Christchurch, New Zealand. Search sent me here. I must have inadvertently damaged it while trying to get home. Thanks for sorting it!

For god's sake. Please look closer and realise it was an accident.

The LINZ buildings weren't there, so it made no sense to continue.

changeset/156559246

All cool. I was originally drawing them by hand too, and quite proud of my results.

Then I found the LINZ data layer, so have been using that. It's not perfect, but it's better than what's there.

I'd recommend you probably stop doing it like that. I think the LINZ stuff is probably good enough, and comes with their building ID, so could potentially be matched and updated as they perfect their system.

I'm working on trying to get the same coverage as other suburbs, and doing a tidyup of everything afterwards.

"Perfect is the enemy of good"

I've found I can manually verify and add several thousand buildings a night.

Thinking about the data, I'm not even sure how much of it is actually useful. Does anyone actually care if a building is a house or a garage? Perhaps a little. But in reality, most likely not.

I think your AEDs are probably an example of the good quality data, even though the out of hospital cardiac arrest survival rates are < 10%, and their value is very questionable. Any idea where the various apps are pulling their data from?

I found a lost of postboxes that I might add later, but they're probably of even more questionable value.

Thanks for that. Yes. I shall be more careful in future. I've been working to try and improve the overall average of the data, trying to do manually supervised mass updates.

Where I can contribute the most with my existing OSM skill level.

I can't believe how many houses are missing, and the massive infill in houses in some really dodgy areas that would never have been allowed under previous rules.

I've spotted at least one future cancer cluster in the making!

As you point out, the time lag in the data is pretty much the issue.

I figured it was machine generated. It's pretty obvious to my eye. It Looooooves the straight lines and 90 degree corners. Container yards get it all excited!

I too have a background in software, radios and paramedicine.

I also have a perfect record in CPR.

They ALL died.

Fuck! Really sorry mate. I hope you can repair my damage.

Genuine question. I'm guessing the buildings are missing from the LINZ imagery I'm using?

Is there a way of easily telling what's been genuinely demolished, vs added manually by someone surveying? I just made the assumption that if the building wasn't in the picture, it'd probably been cleared post quakes.

I realise I have a lot to learn, but genuinely hadn't considered that scenario, even though on reflection, I've added my own survey data.

I'm seeing a lot of empty sections in suburbia with buildings on sections that have been cleared in the photography, which I've taken to mean have been demo'd, and will eventually get added when the images get updated?

Recently, I've been concentrating on adding the tens of thousands of missing buildings from the LINZ data, just to try to improve the average coverage.

From memory, there's no creation date on the buildings I have checked. Or was it just bad luck on the ones I had checked? If I'm missing something, please let me know. I definitely don't want to be recklessly breaking others peoples work

Sorry about the damage, I realise I'm still learning, and are keen to receive help/advice to stop me breaking more than I (think) I'm fixing.

Whoops. Sorry. Yes. You're correct.
I'm happy if you wish to correct it.

Awesome. Thanks Ralley.

Yeah, I'm a newbie at this and it hasn't come out correctly. If someone could tidy it up and make it a cul-de-sac, that'd be great thanks.