Latest version is absolutely decimating my battery

I don’t know what’s up with the app lately but I literally can’t leave my phone unplugged overnight because it’ll be dead by morning. I’m running the latest version of everything on an iPhone 15 Pro Max. I’ve been using Arc for years and I’ve never encountered battery drain like this.

Hi @rolandsmash!

I’m not sure why that would be happening. Nothing’s changed recently in Arc’s underlying recording engine that could plausibly cause that.

There is however a new option in Arc’s Settings → Privacy view, which allows you to toggle on the annoying iOS status bar / Dynamic Island blue “using location” indicator. When that is toggled on, Arc can then use more energy efficient location recording settings, which might help solve the problem you’re seeing overnight.

Aside: That new option is … super weird, on Apple’s part. A while back they made the decision to start terminating apps that make use of the most optimal and efficient all-day location recording settings, unless that app opted in to showing the explicit status bar / Dynamic Island location indicator. It really makes no sense. But it is what it is, so there’s an option now for it in Arc :man_shrugging:t2: Worth a try!

I have that status bar thingy enabled and the app is still draining my battery like crazy. :sob: I love this app too much to stop using it but I wish I knew what Apple changed behind the scenes to create this situation.

I would love to know too.

The change I made to show the blue status bar thing got rid of an inefficiency that Apple’s silly rule brought in, but it doesn’t seem to have solved the core of the problem.

My main phone (iPhone 13 Pro) is also showing considerably higher battery drain when in sleep mode than it used to, and nothing in Arc has changed enough to cause that (that I’m aware of or can find in testing).

Meanwhile my iPhone 11 Pro is doing just fine, lasting for 24+ hours on a single charge.

At this point I’m tempted to file a bug report with Apple. But that step is almost always entirely useless, given that Apple bug reports are black holes, getting little to no feedback from Apple, and no indication that anything is being done or even having the problem genuinely acknowledged.

And that bug report would have to carefully quantify it all, providing exacting test cases, otherwise they’d dismiss it immediately without consideration. But this is a problem that’s inconsistent, happens over long spans of time, has no reliable test cases, etc.

Well, I’m going to keep doing more testing. I’ll either find something helpful, or collate enough evidence to submit some kind of bug report to Apple. (Then probably wait weeks or months for them to not reply and for nothing to be done).

A frustrating problem!

It does feel very random. There are days when I’ve been out a lot, lots of different locations logged in the background, and Arc’s battery usage feels completely normal/nominal. Then there are days where my iPhone has just been sitting on a table somewhere most of the day, unused, and battery usage will be through-the-roof.

iOS 17 has been a mixed bag like that. I remember when it was in public beta, Apple randomly broke the Mail app’s ability to intelligently suggest folders when filing emails. There was never any official documentation that this feature became nonfunctional. It simply stopped working and was quietly removed altogether when iOS 17 was officially released. (It continued to work in Sonoma, which was the only indication that the feature hadn’t been discontinued.) After a couple of iOS maintenance releases, it randomly returned and has been working normally ever since, all without a word of acknowledgement from Apple.

1 Like

Yeah, from the developer perspective that’s the hardest part. 90% of changes (and 99% of bugs) go undocumented and unacknowledged, and there’s no viable direct (or even indirect) line of communication to the relevant Apple teams to confirm, test, or debug. We’re simply left in the dark, with no credible path to resolution of problems on Apple’s side.

Yeah my day to day use of Arc has continued to feel fairly normal on my main phone. Although battery life is going down marginally faster, but the phone is getting older (battery health at 80% now). But then there’ll be times like the other night where I didn’t put the phone on the charger properly, and woke up in the morning to 5% battery, when for that duration of time it should’ve been 50% or more.

There was nothing suspicious in the recorded data, and my energy use profiling of the app the other month when I found that blue status bar thing didn’t show up any other potential pain points.

Hopefully I’ll find some clues during my testing this month. It would be really good to get this one solved.

An update on this: I’m hopeful that Arc 3.16 which went live yesterday should improve the battery life situation. I’m seeing some potentially promising results on my main phone so far.

Unfortunately I have recently updated Arc to 3.16.0 and it was way worse than before.

For example today Arc used 36% of battery:

A week ago using the previous version of Arc, doing the same commute and same things during the day, Arc only used 14% of battery:

It even used less battery than the browsers.

@trackerminerfs At this stage I think iOS’s Battery view is just broken and can’t be trusted at all.

I’ve been getting longer battery life since recent changes in Arc, with noticeably less battery drain, but the iOS Settings Battery view is producing essentially random results each day, with some days showing Arc and Mini using essentially no battery at all while other days it’ll have them at 30-50%!

Then this morning I saw this:

And this:

Which is undeniably nonsense and completely broken. Yesterday I spent over an hour on Threads, but it’s listed down the bottom with no percentage at all. Except just now it jumped up to third position, after opening the Battery view again a few minutes later :man_shrugging:t2:

But the most damning thing is This & That being listed down the bottom with no energy use, and 3 minutes screen time, except that I spent several hours working on it yesterday, with it consuming so much energy during those hours that the phone was in serious thermal state.

At this stage it’s clearly just fake data. I don’t know what’s going on, but it’s unable information.

I think we’re back to just having to observe battery percentage remaining at specific times during the day, and gauging battery life by hours since last charge, while ignoring the Settings Battery view altogether.

I’ve always hated the Settings Battery view because it’s misleading, even when correct. But right now it’s, without exaggeration, just pure fiction. (I’m on iOS 17.3.1).

Oh, Instagram is also listed down the bottom with no energy use, and 10 minutes screen time. Except that I use Instagram an hour or more every day, due to it being my main messaging app.

Yeah, I don’t know what’s going on, but at least in iOS 17.3.1 the Battery view appears to be entirely broken.

I can attest that Arc Mini has also jumped very high in battery usage, as high as 30% one day. Now at 15% Arc Mini and 4% Arc normal.

Yeah Arc Mini is jumping all over the place for me in the Battery view. It’s just nonsense.

In terms of actual real world results, I’ve been getting 8-12 hours from a full charge lately on my main phone (iPhone 13 Pro, with battery health at 79%). That’s with ~5 hours of screen time per day, so it’s actually pretty impressive battery life.

My current theory for what’s going on with the Settings Battery view is: 1) sometimes it’s simply wrong, showing nonsense values, and 2) most days it’s showing Arc and Mini with high percentages due to their scheduled housekeeping tasks that are run while plugged in to power.

Since Arc and Mini migrated to using Core ML for all activity type classifier models, the daily model updating energy cost has gone up. With the old system models were rebuilt per activity type (eg a model fragment for walking, or cycling, etc). So when model updates ran, only the activity types that actually changed got updated. That was much more efficient. Now with Core ML the entire model gets updated each time, so it’s more time and energy expensive.

That would explain why Arc and Mini are showing higher percentages but phone battery life isn’t suffering. Those model updates only happen when the phone is plugged in to power, so they come at no cost to battery life.

I’m toying with the idea of starting on a project of rebuilding LocoKit from the ground up. I’ve been wanting to do it for a long time, as LocoKit gets older and the technologies and coding styles is uses become increasingly out of date. It would be great to do all that low level stuff fresh, taking advantage of the latest tricks and tools. And it would also allow for fine grained energy use testing from the ground up, same as I got to do when LocoKit was first built, years back.

If I do start on that project, I’ll run it as a separate app from Arc and Mini, at least initially, so that I can do careful energy use testing without it getting conflated with UI energy use, daily housekeeping task energy use, etc.

I think a big problem right now is (well aside from Settings Battery view being incredibly untrustworthy) is that there’s a lot of different ways the app could be consuming energy, most of which only happen while plugged in to power, and there’s no way to tell those different energy consumption sources apart.

Right now the only reliable way to determine whether the app is hurting battery life or not is to simply count the hours since last full charge, and calculate percentage lost per hour since then. Basically counting how many hours the phone gets per full charge. That’s really not ideal.

Oh, for anyone wanting to get more useful battery use information than the often-nonsense that Settings Battery view produces: Install Battery Grapher.

It calculates actual useful information, instead of mostly misleading information. And appears to be on average far more correct than Settings Battery view.

For me it estimates 14 hours battery life, which is probably more correct than my on 8-12 estimate. I tend to put it on the charger once it’s around 25% remaining, but that 25% probably has 2-4 more hours in it.