To add to the point: Alphabet has probably most data than any other company (except FB I suppose) and they still can't release a good product that people will actually use no matter how much data they have.
There's this anecdote (which might have been made up entirely to make a point):
Restaurant management wanted to compare different soup offerings by counting orders from each soup to determine which ones were more popular. They have selected the most popular two offerings, the rest were scrapped in order to safe money on ingredients. Soon after, not only did the order numbers of those two soup offerings drop, the total number of soup orders dropped. How come? Well, maybe nobody has asked the customers if the offered soup was tasty at all. A quick survey revealed that customers make the popular choice, find out it's crap, and then do not ever order soup again in that restaurant, or in rarer cases give the other one a shot. It turned out, the most popular offer was basically cheap crap nobody wanted to eat, and when there's nothing else, they keep ordering the same, or never visit that restaurant again.
Telemetry does not tell anything about user preferences. Who ever is selling you that idea, does not either.
Alphabet markets two linux-based operating systems for consumer devices that are extremely popular. Telemetry from the field makes android and chromeos better operating systems.
I don't think the fact that Google has two popular OSes with telemetry means that telemetry makes them better OSes. For Android, their only real competitor is iOS. Their big advantage there is cost and the reason there isn't really another competitor is the difficulty of creating an OS which requires a number of resources. For ChromeOS, the situation is similar with regards to the time to create an OS. I think there, their main competitors are small Linux distributions, but, in addition to manpower, Google has the money to procure and sell laptops with their software already on them at scale. So, I'm not convinced telemetry is actually creating a better product, their products happen to have telemetry.
The is HN so you are allowed to hold forth out of pure ignorance, if you want to. Android-wide and ChromeOS-wide profiling produce binaries including the Linux kernel that are peak-optimized for actual conditions in the field and real use cases. They ship the only profile-optimized Linux kernels you can get from any distribution. It is a demonstrably better product through telemetry.
Got a link to any information on this, especially the collection of the profile information from user's devices? The only reference I can see to PGO on android is the support they have for a more traditional flow (create special instrumented binary, run a 'representative' workload or two, get profile data for PGO). It would be especially interesting if there's any info on how much of a performance improvement this yielded.
Restaurant management wanted to compare different soup offerings by counting orders from each soup to determine which ones were more popular. They have selected the most popular two offerings, the rest were scrapped in order to safe money on ingredients. Soon after, not only did the order numbers of those two soup offerings drop, the total number of soup orders dropped. How come? Well, maybe nobody has asked the customers if the offered soup was tasty at all. A quick survey revealed that customers make the popular choice, find out it's crap, and then do not ever order soup again in that restaurant, or in rarer cases give the other one a shot. It turned out, the most popular offer was basically cheap crap nobody wanted to eat, and when there's nothing else, they keep ordering the same, or never visit that restaurant again.
Telemetry does not tell anything about user preferences. Who ever is selling you that idea, does not either.