One advantage of being a professor is that you can ramble about your eccentric theories to a captive audience. For example, I often grumble to my graduate students that every time a new iPhone comes out, my existing iPhone seems to slow down. How convenient, I might think: Wouldn’t many business owners love to make their old product less useful whenever they released a newer one? When you sell the device and control the operating system, that’s an option.
It should be noted that Mullainathan isn't actually suggesting that there's any merit to the posit above, not that it stops me from being annoyed at the suggestion anyway. For Mullainathan, correlation between searches for “slow iPhone” and iPhone release dates can be explained by Apple's general tendency to release major iOS updates with new phone releases; update your older phone with an iOS meant to take advantage of a new models improved specifications, and maybe your older phone feels a bit slower. That Samsung lacks the correlation between “slow phone” searches and release dates like the iPhone does, can then be explained in that Android kind of sucks at pushing out the newest Android OS versions (let alone updates) to users, so Galaxy owners are less likely to be slowed down by an OS oriented for hardware superior to their own.
I'd add the additional explanation that launch day spikes in searches for “slow Galaxy phone” are likely non-existent because who on Earth even knows when Samsung Galaxy phone release dates are, I mean can anybody recall media coverage of Samsung fanboys camping out to get the newest Galaxy “X”?
Finally, rather than try to determine whether Apple “deliberately” sabotages older models via software (so called “planned obsolescence”) when new phones are released based on search spikes, maybe one could do a search themselves, like say google “iPhone benchmark” or some such, download some benchmark software, and then run some tests to get some actual data, sort of like any number of websites that already do this?