"But Tesla's defense is not that the data is in a proprietary format that can't be decoded."
I never said it was. I'm not sure why you continue to try to come up with unrelated arguments to push your agenda in a discussion completely unrelated to what you're talking about.
"And what do you base this speculation that Tesla does something so strange?"
Because educated speculation can lead to positive and constructive discourse? I have a fair amount of experience in the ML and CS field and I could imagine a situation where the data is in fact difficult to analyze.
Also, logging geolocation data is a fair amount different than logging neural weights or other types of ML data. Without knowledge of the types and composition of ML algorithms those numbers simply won't help you. Which was my point to begin with.
You've proposed a reason for Tesla's reluctance to release data that Tesla itself has not used, so it's you that's bringing up unrelated arguments. Why speculate on the proprietary formats Tesla may or may not use when Tesla doesn't appear to have argued that the data was proprietary?
You're the only one talking about Tesla in this situation. I never mentioned Tesla's reluctance to release data. That's your fabrication.
I'm talking about data that's incapable of being analyzed in a meaningful way without contextual / proprietary knowledge of how that data was created, is used, and is stored. In reality, I'd wager that most companies have more of that type of data than not.
My mistake, I see that your initial post was asking this speculative question. The first response to you shifted the topic to "What does Tesla have to hide?", which seems reasonable given the topic of the submitted article, but that wasn't in your original question.
I never said it was. I'm not sure why you continue to try to come up with unrelated arguments to push your agenda in a discussion completely unrelated to what you're talking about.
"And what do you base this speculation that Tesla does something so strange?"
Because educated speculation can lead to positive and constructive discourse? I have a fair amount of experience in the ML and CS field and I could imagine a situation where the data is in fact difficult to analyze.
Also, logging geolocation data is a fair amount different than logging neural weights or other types of ML data. Without knowledge of the types and composition of ML algorithms those numbers simply won't help you. Which was my point to begin with.