Yeah the stats matter and Linux matters to valve, but OP’s point was that the monthly voluntary surveys have a lot of variance and are of limited value except for trends. So people saying “oh it’s gone up this month” or “down this month” is nonsense and a total lack of understanding of statistics and the nature of the data.
Valve however has much more data to play with - it has the original dataset which matches all the answers together, and can also cross reference it with all the other data it holds for it’s users. So Valve will know much more accurately what is going on; they’d never release the rest of their data as it’s commercially very valuable.
The limited dataset they do release is aggregate data, and from the survey only. It’s interesting but very limited - ok for rough trends, but not much else.
I do agree that monthly voluntary surveys is not an exceptional way for the general audience to have a clear idea on things are going on, but my point is less about the general audience and the responsibility weight Valve itself sit onto.
We can endlessly speculate on the secret/true data Valve is hiding from the general public, but in fact, only the actual public data affect their business: that’s where publisher and developers make the strategic choices on which platform (OS) and hardware (VR HMD, highend GPU…) to support. Any incorrect or nonfactual data would lead to less sales, and less happy developer/publisher/customers.
Yes, of course Valve does have it’s own “secret recipient data” they don’t share… but I think the secret data is used more as sort of control on those who try to cheat the stats.
Sometime simplified Chinese language goes on top, resulting English language as secondary for the whole platform… quite often Valve fix those stats, no doubt by cross referencing their secret sauce.
That’s how most modern anticheat in videogames works: the data is keep secret, until one special day you get one big wave that flush all them at once (if you throw constantly daily updated data on which kind of cheater you caught… the cheater got a precious feedback they can play onto)
Yeah the stats matter and Linux matters to valve, but OP’s point was that the monthly voluntary surveys have a lot of variance and are of limited value except for trends. So people saying “oh it’s gone up this month” or “down this month” is nonsense and a total lack of understanding of statistics and the nature of the data.
Valve however has much more data to play with - it has the original dataset which matches all the answers together, and can also cross reference it with all the other data it holds for it’s users. So Valve will know much more accurately what is going on; they’d never release the rest of their data as it’s commercially very valuable.
The limited dataset they do release is aggregate data, and from the survey only. It’s interesting but very limited - ok for rough trends, but not much else.
I do agree that monthly voluntary surveys is not an exceptional way for the general audience to have a clear idea on things are going on, but my point is less about the general audience and the responsibility weight Valve itself sit onto.
We can endlessly speculate on the secret/true data Valve is hiding from the general public, but in fact, only the actual public data affect their business: that’s where publisher and developers make the strategic choices on which platform (OS) and hardware (VR HMD, highend GPU…) to support. Any incorrect or nonfactual data would lead to less sales, and less happy developer/publisher/customers.
Yes, of course Valve does have it’s own “secret recipient data” they don’t share… but I think the secret data is used more as sort of control on those who try to cheat the stats.
Sometime simplified Chinese language goes on top, resulting English language as secondary for the whole platform… quite often Valve fix those stats, no doubt by cross referencing their secret sauce.
That’s how most modern anticheat in videogames works: the data is keep secret, until one special day you get one big wave that flush all them at once (if you throw constantly daily updated data on which kind of cheater you caught… the cheater got a precious feedback they can play onto)