Respectfully, you can't provide a source and then cherry pick which parts you like and disregard the parts you don't like.
The Geekbench test resulted in worse results than the first test across the board, with times in the 3 hour range v. 4 to 5 hour range.
As stated before, the tests were flawed to begin with. Some things that come to mind are:
- Not using original Apple batteries (preferably that came with the device and have been calibrated for that device in the factory)
- Not addressing the fact of the difference between design capacity and actual capacity, as visible with one example at around 0:49
- Not explaining how they accounted for the differences of actual capacity across batteries
- Some nonsensical thing about "calibrating" the batteries, which not only is a non-existent thing nowadays but is actually not even possible to do based on how iPhones charge and when they decide to shutdown. I can spend an entire week writing up something about this but the gist of it is that the iPhone's power management won't allow you to charge to true 100% nor will it let you deplete it to true 0%, regardless of what percentage is indicated in the upper right.
- No information on if the rest of the iPhone is using genuine parts, the third one from the right, running 12.4.1 stands out because of how different the screen looks compared to the others. This suggests that the screen might not be original
- Similarly the video doesn't mention if these are running on the TSMC chip or the Samsung chip, there have been differences documented in battery performance between the two
- The first iPhone running an older version of Geekbench (4.2.3) than the others (4.4.1), which is not only relevant for the second test, but also in the first test because Geekbench was used a few times to run unrelated benchmarks
- The first test doesn't simulate "actual use" if only because the phones don't ever go into standby, a downtime state the device uses to do some essentials in the background. All it does is give you the illusion that it mimics "actual use", muddying the waters in terms of the precise pressure applied to the battery, in turn making the results unreliable.
- Since this is a test to see which iOS version manages the battery the worst, perhaps a separate tests that allows the iPhone to enter into low power mode can prove to be useful to assess the iOS version's ability to manage the battery when allowed to use the tools it's been given to do so
Overall the methodology for both tests is questionable, but at least the Geekbench test (of the same Geekbench version) should provide a more scientific result just because it should theoretically put equal amounts of pressure on the battery.
That said, and despite the flaws, neither test support the notion that newer iOS versions lead to worse battery lives.
If that were truly the case then the oldest version (10.3.3) would've finished first in both tests and the oldest version (15.3.1) would've finished last in both tests, this didn't happen.
Similarly, 13.7 wouldn't perform better than 12.4.1 in the first test, nor would 15.3.1 perform better than 14.7 in the first test, nor would the third oldest (12.4.1) end up 5th in the second test, etc. etc.
Additionally 15.7 wouldn't perform better in
this video of them, than 15.3.1.
To be clear, I'm not saying that the conclusion (or initial hypothesis) is untrue. I am saying that the evidence in this video doesn't support it.
These contradicting outcomes could very well be caused by the flaws in methodology as laid out above, flaws like that work both ways as they undermine the reliability of the outcomes and the conclusions derived from those outcomes.
Personally I don't recognize your anecdotes, then again it has been a while since I didn't upgrade yearly.
I think there are a few hypothesis that each individually could be true, but I don't hold any knowledge supporting one over the other:
- iOS could be becoming more and more bulkier in terms of (background) functionality, aversely affecting battery life of older devices.
- iOS could become smarter and better at power management and smart decisions about certain functionality in older devices could be made (akin to not enabling the parallax effect on older devices when iOS 7 came out) to prevent extreme discharge of battery on older devices
- Older devices often have a more deteriorated battery and a new iOS version (with or without the effects described in the hypotheses above) can give the erroneous impression that battery life is worsened
Respectfully, you can't provide a source and then cherry pick which parts you like and disregard the parts you don't like.
The Geekbench test resulted in worse results than the first test across the board, with times in the 3 hour range v. 4 to 5 hour range.
As stated before, the tests were flawed to begin with. Some things that come to mind are:
- Not using original Apple batteries (preferably that came with the device and have been calibrated for that device in the factory)
- Not addressing the fact of the difference between design capacity and actual capacity, as visible with one example at around 0:49
- Not explaining how they accounted for the differences of actual capacity across batteries
- Some nonsensical thing about "calibrating" the batteries, which not only is a non-existent thing nowadays but is actually not even possible to do based on how iPhones charge and when they decide to shutdown. I can spend an entire week writing up something about this but the gist of it is that the iPhone's power management won't allow you to charge to true 100% nor will it let you deplete it to true 0%, regardless of what percentage is indicated in the upper right.
- No information on if the rest of the iPhone is using genuine parts, the third one from the right, running 12.4.1 stands out because of how different the screen looks compared to the others. This suggests that the screen might not be original
- Similarly the video doesn't mention if these are running on the TSMC chip or the Samsung chip, there have been differences documented in battery performance between the two
- The first iPhone running an older version of Geekbench (4.2.3) than the others (4.4.1), which is not only relevant for the second test, but also in the first test because Geekbench was used a few times to run unrelated benchmarks
- The first test doesn't simulate "actual use" if only because the phones don't ever go into standby, a downtime state the device uses to do some essentials in the background. All it does is give you the illusion that it mimics "actual use", muddying the waters in terms of the precise pressure applied to the battery, in turn making the results unreliable.
- Since this is a test to see which iOS version manages the battery the worst, perhaps a separate tests that allows the iPhone to enter into low power mode can prove to be useful to assess the iOS version's ability to manage the battery when allowed to use the tools it's been given to do so
Overall the methodology for both tests is questionable, but at least the Geekbench test (of the same Geekbench version) should provide a more scientific result just because it should theoretically put equal amounts of pressure on the battery.
That said, and despite the flaws, neither test support the notion that newer iOS versions lead to worse battery lives.
If that were truly the case then the oldest version (10.3.3) would've finished first in both tests and the oldest version (15.3.1) would've finished last in both tests, this didn't happen.
Similarly, 13.7 wouldn't perform better than 12.4.1 in the first test, nor would 15.3.1 perform better than 14.7 in the first test, nor would the third oldest (12.4.1) end up 5th in the second test, etc. etc.
Additionally 15.7 wouldn't perform better in
this video of them, than 15.3.1.
To be clear, I'm not saying that the conclusion (or initial hypothesis) is untrue. I am saying that the evidence in this video doesn't support it.
These contradicting outcomes could very well be caused by the flaws in methodology as laid out above, flaws like that work both ways as they undermine the reliability of the outcomes and the conclusions derived from those outcomes.
Personally I don't recognize your anecdotes, then again it has been a while since I didn't upgrade yearly.
I think there are a few hypothesis that each individually could be true, but I don't hold any knowledge supporting one over the other:
- iOS could be becoming more and more bulkier in terms of (background) functionality, aversely affecting battery life of older devices.
- iOS could become smarter and better at power management and smart decisions about certain functionality in older devices could be made (akin to not enabling the parallax effect on older devices when iOS 7 came out) to prevent extreme discharge of battery on older devices
- Older devices often have a more deteriorated battery and a new iOS version (with or without the effects described in the hypotheses above) can give the erroneous impression that battery life is worsen
iOS 10 was better than iOS 15, both times. It‘s not 10 > 11> 12… etc. It’s 9/10 > 14/15, regardless of what happens in the middle. That’s all that matters. It doesn't matter if 13 is slightly better than 12, because 13 is worse than 10.
Using original Apple batteries wouldn’t change anything.
Differences in battery capacity between actual and stated are often negligible.
I agree, the calibration part is nonsense exactly because of what you said.
I don’t know about original parts. It shouldn‘t change much, regardless.
Agree on TSMC vs Samsung, it shouldn’t change the end result much, regardless.
Disagree with your assessment on reliability. You try to test real-world conditions, not lab conditions, as those are irrelevant and frequently differ from real-world use. (Millions of examples about this, I can give Apple’s stated runtimes as one: I’ve been able to match those under extremely light use, and oftentimes I can’t even match them. Apple claims 10 hours of web browsing on LTE for the iPhone 6s. I have been able to reach that exactly zero times, on any version of iOS, with any battery condition. Why? Because the real world isn’t a lab. Signal quality varies, the distance to the tower varies, alongside the probably hundreds of factors which determine battery life. It is impossible to match everything). Is the test perfect? No. Is it reliable enough? Generally, yes.
Also disagree on Low Power Mode, I’ve tested three iOS versions with the same settings and usage, with and without Low Power Mode, newer versions are always worse.
Regarding your hypothesis:
Agree with the first one. I cannot test it, but that has been my running theory for ages. I recall somebody mentioning that background processes on the 6s were a lot more numerous on iOS 15 compared to iOS 10, but I can’t cite it, sorry.
Agree with the second one, but due to the first one, I reckon it doesn’t make much difference.
Disagree with the third one, because whilst that is true, replacing the battery fails to match original versions of iOS (even if it improves it, however significant that improvement might be).
Respectfully, I can’t help but comment on this, sorry. Why is this always the case? Why does everyone who debate this upgrade every year? Why? Of course, if you upgrade every year you‘ll never notice it. There‘s a reason people who use older iPhones for years always say this, because regardless of better or worse tests, regardless of Geekbench, regardless of differences between people’s settings, battery life for the same device is always worse when updating. Is the test perfect? No! Because of everything you said, varying battery health, varying usage and settings, altered usage patterns, all of that. It‘s not Lab-like. But it never fails. On real-world conditions, it is always worse. (Like the first test shows, regardless of its imperfection).
Edit: you mentioned standby, and standby time decreasing has been a staple complaint of 6s users since at least iOS 13. I reckon it would make it worse, not better. In fact, many have said that while screen-on time decreased considerably when compared to iOS 10 (like tests show, again), they were more bothered by standby decrease, which precludes even using the device less to conserve battery (which, for an updated device with not great battery health -i.e, the vast majority of 6s devices running around - battery life is degraded enough to make it unusable even for a very light user).
