Slashdot Mirror
Do Apple and Google Sabotage Older Phones? What the Graphs Don't Show
Harvard economics professor Sendhil Mullainathan takes a look in the New York Times at interesting correlations between the release dates of new phones and OSes and search queries that indicate frustration with the speed of the phones that people already have. Mullainathan illustrates with graphs (and gives plausible explanations for the difference) just how different the curves are over time for the search terms "iPhone slow" and "Samsung Galaxy slow." It's easy to see with the iPhone graph especially how it could seem to users that Apple has intentionally slowed down older phones to nudge them toward upgrading. While he's careful not to rule out intentional slowing of older phone models (that's possible, after all), Mullainathan cites several factors that mean there's no need to believe in a phone-slowing conspiracy, and at least two big reasons (reputation, liability) for companies — Apple, Google, and cellphone manufacturers like Samsung — not to take part in one. He points out various wrinkles in what the data could really indicate, including genuine but innocent slowdowns caused by optimizing for newer hardware. It's an interesting look at the difference between having mere statistics, no matter how rigorously gathered, and knowing quite what they mean.
It seems like *everything* starts slowing down or breaking for no reason. I don't buy wear/tear as a reason when everyone and their grandmother suggests that you need to update the firmware to get it working again. If it worked fine with the old firmware, why is updating the firmware fixing the problem? WHAT DID YOU DO TO MY OLD FIRMWARE!?
speed is about the same too. i plan to use it until it dies or i can't get any new apps which will probably be a year or so after ios 9 ships
Apple, who continues to provide updates to older phones years after they're released. Some users find these new features slow down their older phones.
Google, who doesn't have a mechanism to provide updates to almost any of their phones years after they're released. Some users find their phones slow down years after release anyway.
It is a big leap, there could be various other explanations of varying degrees of malice. As the new release comes through, bug fixes for older releases are put on back burner, apps are changed and tuned to take advantage of new version run slower in older version.. Or the way graphics subsystem is organized in iOS might have different bottlenecks based on the display resolution. So as new releases come in, default sizes for buffers and hashtables might change deep in the OS slowing down older apps.
And if you are going to postulate "Apple might slow down older versions deliberately", why can't you postulate, "Google might spike and skew the history of the past searches to make Apple look bad"?
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
Interesting musings but the first thing that came to my mind is the reverse - is Google sabotaging the search results? I know this sounds a bit strange but could it be possible that Google is being 'creative' with the raw Google Analytics data. Would it not serve Google's best interest to fudge the results to make Apple look bad right around the time of one of their releases trying to drive people to Android.
My $0.02
I don't think there's any reason to read a conspiracy into the situation. They release a new phone that's much faster, and then they release an updated OS with new features to take advantage of that extra computing power. Adding features that use more computing power makes the old phones seem slow.
I'm tempted to compare it to the development of desktops and laptops, both of which went through similar upgrade cycles before leveling off a bit. However, there's a big difference in that desktop and laptops were developing quickly to cram features into the OS, at the cost of focus on efficiency, which serves as a partial explanation as to why things became "slow" with upgrades. Desktop and laptop software went through a period of bloat, and then in recent years, additional features traded off against speed gains from recoding things with efficiency as a goal. Meanwhile, Android and iOS needed to be written to be efficient from the start. They wanted to make the hardware as small/thin/light as possible, which meant that the power requirements had to be low. To give an example of the effects of this, a requirement for using as little power as possible has been the reason iOS has always limited multitasking.
It is a bit strange they did not correlate to iOS releases, but iPhone releases.
I find it much more likely that when you upgrade to iOS+1, the new features slow your phone down. I've experienced that several times, my 3GS became "much slower" after upgrading it. The new iOS had more eye candy etc.
But that's not the same as saying the old hardware is slower.
Probable impossibilities are to be preferred to improbable possibilities.
Aristotele
As the author points out, each phone release is accompanied by a major OS release. With a major software release comes bugs, as well as a raft of CPU-eating new features to play with, so it makes perfect sense that there would be a spike in complaints about performance and a host of other issues. No conspiracy necessary.
How can I believe you when you tell me what I don't want to hear?
The article says, "phones feel slower over time as they hold more software". How does this follow? How does the phone get "slower" just because more software is installed? This sounds an awful lot like the cargo cult thinking of "well the hard drive is full so we have to buy a new computer because this one is slow."
I know some software will start agents on boot, but they just sit in the background and do little. top reveals very little CPU time and memory consumed by these.
Shutting down free speech with violence isn't fighting fascism. It IS fascism!
The methodology of testing the hypothesis is to look for google searches about "iphone slow" or "samsung slow". Assumption made is if people search for "iphone slow" Apple might have done something to slow down iPhones. The control group is Samsung which has the same motive as Apple but not the means because it does not control the OS.
Actually, the data was gathered to see if the professor's view that his phone had slowed down was also shared by other iPhone users; they found an interesting correlation between search spikes and new iPhone models but were careful to say that doesn't prove anything other than people perceive a slowdown when a new phone comes out. He points out some valid reasons why the Samsung / Apple data differs, primarily that Apple releases a new version of IOS with the new iPhone and thus the new iOS may not be optimized for older hardware while many Android users remain on an older version. In addition, since the Andriod device makers don't control Android they may find it cheaper not to spend a lot of time on the OS and rather invest in hardware improvements as the differentiator.
I'm a consultant - I convert gibberish into cash-flow.
Let me guess it doesn't support all the 4G frequencies modern phones do.
My samsung epic 2 gps antennae is much weaker than it used to be.
I suspect the other hardware is also designed to be "good enough to last a few years but not a decade" to save a few pennies.
She was like chocolate when she drank... semi-sweet at first and then increasingly bitter.
The methodology of testing the hypothesis is to look for google searches about "iphone slow" or "samsung slow". Assumption made is if people search for "iphone slow" Apple might have done something to slow down iPhones. The control group is Samsung which has the same motive as Apple but not the means because it does not control the OS.
It is a big leap, there could be various other explanations of varying degrees of malice. As the new release comes through, bug fixes for older releases are put on back burner, apps are changed and tuned to take advantage of new version run slower in older version.. Or the way graphics subsystem is organized in iOS might have different bottlenecks based on the display resolution. So as new releases come in, default sizes for buffers and hashtables might change deep in the OS slowing down older apps.
And if you are going to postulate "Apple might slow down older versions deliberately", why can't you postulate, "Google might spike and skew the history of the past searches to make Apple look bad"?
There's another problem with his theory as well; as we all know, Android phones don't get many OS updates, if any at all. Every study that checks (using real methods) the Android versions currently in use based on hardware, vendor, or general population finds that unless you bought your phone very recently, there's almost no chance you're running the latest version of Android. So how is it that Google is managing to slow down old phones with code in the new versions of Android in the first place?
For your security, this post has been encrypted with ROT-13, twice.
More and more other people have 4G phones using those frequencies, is my best guess.
The concept is called planned obsolescence , and it has existed for as long as people have been buying things.
In the end, the professor, after writing thousands of words, comes to no conclusion.
He's an economist. That's his job.
I'm a consultant - I convert gibberish into cash-flow.
Studies have also shown that as humans age their rate of recall also slows down, not because their brains are slower but because they have to navigate a database filled with entangled excess information. I've noticed that google searches by voice are vastly more word-accurate than siri searches by voice. But that's because google is doing something in the context of something else-- it has clues to context. Siri is trying to do free-form semantics over a much greater realm of possibilities. When you narrow Siri to a phone specific function, it does better than google. As the AI realm grows, perhaps to include sarcasm and slang, these services will require even more compute power to keep going.
However, these days, phone services are done on back end servers, so there is no great reason they should slow down in "modern" times.
Some drink at the fountain of knowledge. Others just gargle.
If you plan to support new code base on old devices at all, development of a large project will result in hundreds of decision points where you can either have more features and faster or easier to maintain code on shipping hardware or better performance on discontinued devices. Just how much effort would YOU spend in the later, especially with a hard deadline coming up?
A new OS is also likely to create new demands on device drivers. How much support are you going to get from the manufacturers after they have discontinued the hardware, got out of an entire area of business or simply went belly up? Anyone who has a working knowledge of the chipset could already have left the company or be engaged on other pressing projects.
I think the most realistic solution is to release all available and legally unconstrained knowledge about the platform to community so that they can provide solutions like CyanogenMod as long as there is sufficient interest. In the meantime, try to treat free updates to discontinued hardware as a glass half full. The vendor has spent millions of dollars developing, testing and certifying it, with no commercial gains for itself besides reputation.
Hasn't Google been moving more things under the umbrella of their much more restrictively licensed Google Play Services? Basically building much of the face of Android on things no longer/never part of AOSP?
fencepost
just a little off
I don't think they intentionally sabotage them, but they are definitely slower. It's just that they keep getting pushed new OS updates, and new app updates, and the new updates expect faster hardware.
I bought a Galaxy Nexus 2 years ago, and when I got it everything was blazing fast. By a few months ago, my phone was frustratingly slow. For a while I considered that I was just looking at newer phones and thinking "Wow, that's so much faster than mine" but just thought that it was the comparison that made mine feel slow, not that it really was slow. I finally decided my phone was objectively slow, not just by comparison.
A few weeks ago I 'downgraded' to Android 4.2.2 (Had been running Android 4.3), and turned off auto app update. (I had previously tried various ROMs with 4.3, but they were all still slow.) Now my phone is fast again. Maybe not as blazing fast as it was brand new, but I no longer feel like I'm ever waiting for the phone.
So, I don't think there is a real 'conspiracy' to slow down old phones, but I think that old phones *are* slower, they know they are doing it, and just don't care. Why would they? They think we should be happy they are pushing updates to us, but they don't think they should have to worry about the experience on devices they sold years ago. But, I have the ability to refuse these updates, so I can keep my personal phone usable for the foreseeable future.
So you're bragging about Google improving the very bad Dakvik engine after 7 years and you're comparing that to iOS which has been completely compiled to native code since day?
Google has been promising 60fps animation for 5 years -- something Apple did in 2007.
Exactly! The methodology is incorrect. And, after having spoken with the good people at AT&T (that's right, buy at the sign of the Death Star) it is the Telcos that are responsible for slow-downs, not the telephone makers.
Why? The Telcos want you using the latest tech so that you will have a two-year contract with them that you cannot easily get out of without paying them lots of money. This keeps you "loyal." And it gets you on the treadmill of upgrades that ensures your loyalty. So what the telcos do is that they "sunset" technology that supports the older phones. And all of their upgrades on their cell towers (which usually aren't really towers that much any more) support new radios and signaling, not the old stuff.
So blame Apple and Samsung all you want, but it's the Telcos that are responsible for slowing down the older tech, not the manufacturers.
Gods don't kill people, people with gods kill people.
Hypothesis one: users' irrational fetish/honeymoon-glow feeling about their old phone is brainwashed away by the advertising for the new phone model so that they're better able to perceive its slowness relative to laptops that cost the same amount.
Hypothesis two: users aren't searching because they suddenly think their phone is slow. They always thought their phone was slow, but before there was nothing they could do about it. They're searching for reviews comparing the speed of new and old phones, to see if there's really an improvement.
Hypothesis three: Apple and Google are somehow slowing down the phones without pushing software updates to them (since, except for Nexus, Google isn't even _able_ to time software update pushes).
Possible journalist approaches: measure the speed of the fucking phones. Ask users why they were searching with a poll on the landing page for the search. Conduct a web poll trying to distinguish hypotheses one and two, shortly after the release of a new phone. Ask Google and Apple if they are slowing phones down, and if they have any historical phone speed measurements they would like to share so you don't have to do it yourself.
To write the actual article: delete hypotheses one and two, then say "I have no evidence to confirm or deny [complicated hypothesis three scenario] but let me explain how it would work."
Fuck this guy. Google and Apple deserve, and desperately need, better criticism than this.
Even a perceived slowdown isn't a good metric as it's terribly subjective. I have an older computer that still runs just as well as when I purchased it, but it feels a lot slower since my new computer has an SSD and once you experience that, you can't go back without feeling like everything is painfully slow, never mind the extra cores and additional RAM. It was the same with dial-up internet back in the day. It didn't get any slower, but once you had used a cable or T1 connection, loading webpages on a 56k connection felt like a small eternity.
If someone were actually interested in evaluating this, they should buy some new phones and benchmark them for several different tasks and then wait a few years to compare the results after a few operating system upgrades. There are probably a few people who have never upgraded their devices, so even today a comparison could be made.