10 ms·

Opt-in probably means you‘ll never get a good sample size. It has to be representative. If it’s 5% there’s a risk of it being unrepresentative even if it’s a la

Opt-in probably means you‘ll never get a good sample size. It has to be representative. If it’s 5% there’s a risk of it being unrepresentative even if it’s a large sample. The risk is a particular type of user or users from a particular region, skill level or age group opting in.

seanhunter 2y ago
Well that's a risk the developers just have to take. It's not even remotely close to being worth sacrificing their entire user base's privacy for. It blows my mind how far the debate has shifted that people think ubiquitous telemetry is even slightly reasonable just because the devs need representative sample of metrics. They really don't. They will be just fine without any metrics at all.
- alkonaut 2y ago
  > sacrificing their entire user base's privacy for. I keep asking what the privacy issue is in anonymous usage statistics, but there are never any answers. Remember: The "anonymous" in "anonymous usage statistics" means there can't be a privacy issue. Then it's not anonymous any more! You could think of it as for example, sending only the program version in, and the only stored data is how many sessions are run of each version, every month. So you could know "is it reasonable to deprecate version 1.0 and only support 2.0?". What exactly is it that you think is an invasion of privacy here? The fact that it makes http requests? > They will be just fine without any metrics at all. Most who added telemetry has a time from before metrics and a time after. And I imagine very few would want to go back. It's driving blind. You can't simply "ask your users". They don't know whether they use Ctrl+C or use Copy on the context menu. And if you ask them, you find that they think they did one thing when in reality they didn't. Knowing what they actually do instead of what they think they do or say they do is invaluable. And luckily at least in enterprise, users are pretty happy to supply this, but it's definitely based on a trust that already needs to exist between seller and customer.

seanhunter 2y ago

Well that's a risk the developers just have to take. It's not even remotely close to being worth sacrificing their entire user base's privacy for. It blows my mind how far the debate has shifted that people think ubiquitous telemetry is even slightly reasonable just because the devs need representative sample of metrics. They really don't. They will be just fine without any metrics at all.

alkonaut 2y ago
> sacrificing their entire user base's privacy for. I keep asking what the privacy issue is in anonymous usage statistics, but there are never any answers. Remember: The "anonymous" in "anonymous usage statistics" means there can't be a privacy issue. Then it's not anonymous any more! You could think of it as for example, sending only the program version in, and the only stored data is how many sessions are run of each version, every month. So you could know "is it reasonable to deprecate version 1.0 and only support 2.0?". What exactly is it that you think is an invasion of privacy here? The fact that it makes http requests? > They will be just fine without any metrics at all. Most who added telemetry has a time from before metrics and a time after. And I imagine very few would want to go back. It's driving blind. You can't simply "ask your users". They don't know whether they use Ctrl+C or use Copy on the context menu. And if you ask them, you find that they think they did one thing when in reality they didn't. Knowing what they actually do instead of what they think they do or say they do is invaluable. And luckily at least in enterprise, users are pretty happy to supply this, but it's definitely based on a trust that already needs to exist between seller and customer.