5 ms·

> This will allow us to track which installs result from which downloads to determine the answers to questions like, "Why do we see so many installs per day, bu

> This will allow us to track which installs result from which downloads to determine the answers to questions like, "Why do we see so many installs per day, but not that many downloads per day?" What value does Mozilla see in being able to do that?

cxr 5y ago
The prevailing belief in the industry is that any problem can be solved with more data, dumpster-grade[1] though it may be. It's appealing to think that it can be used in lieu of just making thoughtful decisions. 1. https://news.ycombinator.com/item?id=25016532 https://news.ycombinator.com/item?id=25016532
ahmedfromtunis 5y ago
I have no idea what the actual answer is, but one can image that they want to understand where are people getting these copies of ff if they do not download them themselves. Why would this be important, again just speculating: to try to leverage whatever channel this might uncover to distribute even more copies of the browser. Is this ID thing the best way to do it, though? Probably.
- noAnswer 5y ago
  > but one can image that they want to understand where are people getting these copies of ff if they do not download them themselves. These people are getting it obviously from their admins! (Like myself. I push Firefox updates to close to 1000 PCs. I thought (amongst other things) I'm doing them a favour by saving them traffic.) They obviously know that! The real reason they are doing this is simply because they started collecting data. Now they are hooked and constantly want more. That is all there is to it. They already identify each individual installation, so someone on the team said: Let's identify each download too.
  - detaro 5y ago
    > These people are getting it obviously from their admins! That's one possibility. And even if you pretend it's the only one, it's still interesting how that's distributed. Is it X admins of 1000-PC orgs, or X*500 people who also install it on dads PC? Count how often each token is used, and you got a histogram, and if you just record that you've not collected anything remotely sensitive about any user. This question can be answered extremely cleanly.
    - noAnswer 5y ago
      I'm an admin of Y orgs + a hand full of private (dad) PCs. The information is useless. They collect it because they can. (Well technically I also push a installation profile that deactivates all known telemetry. So hopefully they get nothing.) > This question can be answered extremely cleanly. What was the question again?
  - GekkePrutser 5y ago
    I pushed it to tens of thousands of endpoints in the past :P Though we have long since moved it to optional. First we moved to Chrome as the "standard browser" due to user requests and eventually our management caved in to Microsoft's constant pushing to make Edge the standard browser. Not that I cared because both are bad for privacy obviously. But the amount of lobbying for edge that they do makes me really sceptical. It's clearly not about having their pet project succeed, they must have some serious strategy hinging on this. Edge doesn't make them any money on its own as it's a free product so the revenue must come from side channels of its use instead. Which, considering MS' past with IE and their recent ventures into tracking and advertising in the OS is probably bad news for the end user :(
- foerbert 5y ago
  Does this even help with that though? Without some further means of identifying users, what use is there in saying download#123 got installed x times? Even if you add the obvious IP information to this, then what? Run GeoIP and say "oh, interesting" when they do or do not correlate? What could they realistically figure out from this that could help them figure out how people are getting Firefox?
Barrin92 5y ago
the reverse case might be more interesting. Many downloads but few install follow throughs may suggest the installation process is to cumbersome or something along those lines.
rtpg 5y ago
That's an easy one, of course. By identifying a "commonly-used installer" (say, some installer stored by some IT team) you could then do things like say with more certainty "no, we do need older installers to keep working". Or reach out to mirrors that have decided to stop updating installers. From an Ops perspective, having an idea of how your install base is installing your browser can have a lot of implications. And not having the information could easily lead you to accidentally breaking installations for large subsets of your users. There's also content-market-y stuff involved, of course.
kingcharles 5y ago
Does the data even answer that question? All it does is separate out two numbers: downloads and installs. How does that tell you why someone is installing a specific download on a specific day?

cxr 5y ago

The prevailing belief in the industry is that any problem can be solved with more data, dumpster-grade[1] though it may be. It's appealing to think that it can be used in lieu of just making thoughtful decisions. 1. https://news.ycombinator.com/item?id=25016532 https://news.ycombinator.com/item?id=25016532

ahmedfromtunis 5y ago

I have no idea what the actual answer is, but one can image that they want to understand where are people getting these copies of ff if they do not download them themselves. Why would this be important, again just speculating: to try to leverage whatever channel this might uncover to distribute even more copies of the browser. Is this ID thing the best way to do it, though? Probably.

noAnswer 5y ago
> but one can image that they want to understand where are people getting these copies of ff if they do not download them themselves. These people are getting it obviously from their admins! (Like myself. I push Firefox updates to close to 1000 PCs. I thought (amongst other things) I'm doing them a favour by saving them traffic.) They obviously know that! The real reason they are doing this is simply because they started collecting data. Now they are hooked and constantly want more. That is all there is to it. They already identify each individual installation, so someone on the team said: Let's identify each download too.
- detaro 5y ago
  > These people are getting it obviously from their admins! That's one possibility. And even if you pretend it's the only one, it's still interesting how that's distributed. Is it X admins of 1000-PC orgs, or X*500 people who also install it on dads PC? Count how often each token is used, and you got a histogram, and if you just record that you've not collected anything remotely sensitive about any user. This question can be answered extremely cleanly.
  - noAnswer 5y ago
    I'm an admin of Y orgs + a hand full of private (dad) PCs. The information is useless. They collect it because they can. (Well technically I also push a installation profile that deactivates all known telemetry. So hopefully they get nothing.) > This question can be answered extremely cleanly. What was the question again?
- GekkePrutser 5y ago
  I pushed it to tens of thousands of endpoints in the past :P Though we have long since moved it to optional. First we moved to Chrome as the "standard browser" due to user requests and eventually our management caved in to Microsoft's constant pushing to make Edge the standard browser. Not that I cared because both are bad for privacy obviously. But the amount of lobbying for edge that they do makes me really sceptical. It's clearly not about having their pet project succeed, they must have some serious strategy hinging on this. Edge doesn't make them any money on its own as it's a free product so the revenue must come from side channels of its use instead. Which, considering MS' past with IE and their recent ventures into tracking and advertising in the OS is probably bad news for the end user :(
foerbert 5y ago
Does this even help with that though? Without some further means of identifying users, what use is there in saying download#123 got installed x times? Even if you add the obvious IP information to this, then what? Run GeoIP and say "oh, interesting" when they do or do not correlate? What could they realistically figure out from this that could help them figure out how people are getting Firefox?

Barrin92 5y ago

the reverse case might be more interesting. Many downloads but few install follow throughs may suggest the installation process is to cumbersome or something along those lines.

rtpg 5y ago

That's an easy one, of course. By identifying a "commonly-used installer" (say, some installer stored by some IT team) you could then do things like say with more certainty "no, we do need older installers to keep working". Or reach out to mirrors that have decided to stop updating installers. From an Ops perspective, having an idea of how your install base is installing your browser can have a lot of implications. And not having the information could easily lead you to accidentally breaking installations for large subsets of your users. There's also content-market-y stuff involved, of course.

kingcharles 5y ago

Does the data even answer that question? All it does is separate out two numbers: downloads and installs. How does that tell you why someone is installing a specific download on a specific day?