I saw today that my light rail trip had two different portions where the trip line randomly snapped off of the tracks and ran along the parallel roads before snapping back. I’ll attach photos but is that Arc or the GPS doing that?
Could also just be a weird coincidence
That’s very strange! Arc / LocoKit2 don’t do any road snapping. In practice it’s very rarely helpful, and almost always just makes things worse.
Though in one of the iOS 26 (or possible iOS 18) betas I did see iOS doing road snapping on its side, sporadically. Which was a worry, because there’s no way for apps to opt out of that.
Which iOS version are you on? I just updated to 26.4 this morning myself. I wonder if Apple are experimenting with doing automatic road snapping in 26.4. Though I really hope not! But those screenshots do look very much like road snapping 
I haven’t updated yet so I’m still on 26.3.1
I’m home now and went back to zoom the rest of the day. I found two more suspicious lines.
Probably irrelevant but I’m in the NYC metro. I don’t know if Apple ever uses us as guinea pigs
1 Like
Yeah those both look highly suspicious. If it’s not snapping to roads, it’s managing to coincidentally look very much like it.
Hmm. Not sure what we can do about that. Road snapping isn’t something Apple expose as a Core Location feature. The closest might be the manager’s “activityType”, which can be set to automotiveNavigation, fitness, otherNavigation, airborne, or other.
None of the explicit ones are suitable for Arc, which is an everything app. So I leave it on the default “other”. Perhaps iOS is foolishly trying to self adjust it and choosing automotiveNavigation*.*
But given we’re already told it we want “other”… there’s nothing else we can do there.
Hmm. At this point I guess we just have to hope it’s short lived, and goes away on its own. If it sticks around I might have to escalate it to Apple.
1 Like
For what it’s worth, I too experienced this on a train last week.
I watched it happen in real-time. Every time the track crossed or got near to a road, my current position indicator would jump to the road. It would then try to follow the road – with a larger and larger “precision” ring – until it got so far away that it could no longer keep pretending.
Unsurprisingly, Google Maps showed the same thing.
1 Like
Most curious!
It sounds like Apple are experimenting with some auto logic for sometimes engaging snap-to-roads. Not always on, but some heuristics where Core Location thinks it might be the right answer, so engages it for a bit.
Not great for Arc! But given there’s no setting for this, no way to opt out of it, unfortunately still nothing we can do about it.
If it starts to become more common, more aggressive, I think we could file a bug report with Apple about it. But for those you tend to have to provide an exact reproduction case, otherwise they will ignore it for the next 10 years. For this one there’s no way for us to reliably reproduce the problem I think, so we won’t achieve much in reporting it.
Anyway, let’s keep an eye on it in the meantime. Fingers crossed it stays an occasional curiosity and minor annoyance and not a frequent data mangling.
This doesn’t seem to be new; I can find instances of people complaining about it online for at least 10 years.
Interestingly, this guy found that “otherNavigation” seemed to be better than “other” (albeit 3+ years ago): https://stackoverflow.com/a/73909446
I can probably reproduce it in the iOS simulator, but if it’s going on 10+ years, I can’t see it getting changed.
I haven’t been replying to this thread because I felt like it’d be redundant, but I’ve loosely been monitoring it and it happens pretty regularly and sometimes pretty consistently.
Idk if this would be compelling as a repeatable bug but it’s probably about a third of my trips through this spot.
EDIT: Here’s the actual line for reference. Elevated rail line without a many tall buildings around. You can see that it runs parallel with the highway for awhile which likely causes the confusion
That’s super interesting, and new info! I hadn’t seen that one before.
For the 10 years, that’s also news to me. I think the first I heard of it (or saw of it) might’ve been only 3-4 years ago at most.
I guess the question is whether it’s becoming more a more frequent thing or not. And @twilldre’s reply suggests that for some people it’s actually quite frequent.
I’m thinking we should try the otherNavigation setting experimentally, perhaps in a TestFlight build or some such (TBD on shape of experiment). It could have other undesirable side effects, so we wouldn’t want to just switch it over then ship it to all. Feel it out first.
We’ve filed BIG-543 for that, so it’s logged now.
Ok that makes it super clear. Unambiguous that in your first screenshot it’s snapping to roads repeatedly for those trips. Damn.
So sounds like at least for some people this is actually a quite common problem. Which means the BIG-543 experiment mentioned above is something we should get to trying soon!
Another source for otherNavigation being the solution, from 10 years ago: Jeffrey Friedl's Blog » More Details on the Insidious iOS Snap-to-Road “Feature” (and a previous entry in which he did a fair amount of testing Jeffrey Friedl's Blog » The Scourge (or Beauty) of “Snap To Road” with iPhone Location-Tracking Apps)
Unfortunately he also says that if any other app is running, in any other mode, road snapping is the likely result. (sigh)
1 Like
Yeah it’s interesting you haven’t heard much from other people. These three instances are all just from this morning on a 45 minute round trip. I could probably reliably find a couple on a majority of my train trips. I wonder what is making me more susceptible.
Probably irrelevant but iPhone 16e on 26.4.2, with significant locations and routes flicked on. Just thinking about @kronk mentioning other apps interacting with location.
Not the end of the world. Most of the snaps are small enough that I’d never notice unless I was looking, but I’m curious.
1 Like
Useful links @kronk! Thanks for those.
Yeah that’s… sigh indeed. Though I wonder if that’s changed over the years.
It used to be that it seemed like all apps were getting the same location results, with one central system curating the next location update that all apps would get. But in more recent years I’ve found that different apps can be getting served completely different location data. Like one app getting high accuracy data while another is getting given quite terrible data at the same time (and not due to the Precise Location setting).
Though for snap to roads perhaps it’s a different deal. We’ll see.
Yeah in one sense it’s no big deal. But in another… this is our long term recording, our personal timeline histories. It would be nice if it weren’t being subtly or significantly messed up each day, and unnecessarily so. Apple’s system thinks it’s being helpful, when it really isn’t.
Hopefully the experiment will play out well! I might get into it while I’m in Bangkok this next two weeks, because I’ll also be doing underground train testing, to try to improve that stuff too.
Just to add on so you have a better picture. This is my 2025 yearly in the same spot compared to the one I posted earlier. I probably did close to double the amount of trips though here than I did so far this year, but 25 is still cleaner (still a few major snaps).
I think you’re right that it’s something Apple is messing with, but not fully brand new it seems. I also swapped from an iphone 12 to 16e midway through the year if that makes a difference. Thanks for looking into this stuff!
@twilldre I wonder if for 2025 it was a case of old LocoKit doing more aggressive smoothing, while LocoKit2 in AT4 has a smarter Kalman that more intelligently filters, retaining more detail while still denoising.
Though hard to be sure on that. Could easily be that Apple are fiddling the snap to roads thresholds over different iOS releases or device models. In the past the strongest signal I’ve got on the issue has been during iOS beta periods - people reporting it when they’re on beta builds.
But that beta builds effect could be iOS updates flushing Core Location’s caches (wifi hotspots, other things), which would result in it resetting and starting fresh on “learning”. That’s a definite effect that I see with the wifi hotspot triangulation, going briefly back to zero after major iOS updates and during the beta periods. Though whether snap to roads decisions play into that also… hard to guess.