"Historically, the survey used a client-side method to ensure that systems were counted only once per year, in order to provide an accurate picture of the entire Steam user population. It turns out, however, that many cyber cafes manage their hardware in a way that was causing their customers to be over counted. [...] It took us some time to root-cause the problem and deploy a fix, but we are confident that, as of April 2018, the Steam Hardware Survey is no longer over counting users."
The Fix doesn't make it accurate because how it must be done is assigning an ID to each steam installation or the system installation of the user.
But what if the user resets his Windows 10 to factory defaults and starts over installing Steam again because an broken Windows update killed his installation? -> Then the ID should also be counted as a new one.
What if a Linux user just changes his package sources to upgrade and never formats his computer just because it runs? He can only get the survey exactly 1 time per year.
But what if you run Steam in Wine for your 50 games with 50 wine versions? Don't the installations count as 50 different ones with 50 IDs?
What if you run a VM, with GPU passthrough and copy the virtual hard drive 20 times before you install Steam?
What if you get a brand new system image from the network each time you start your computer?
Some of these issues could be fixed perhaps but i am absolutely sure that you would only get accurate numbers when you only allow every Steam account (Edit: that has to exist for some time, e.g. one month and has to have a specific amount of value in the library) to do the survey on a client of their choice a single time per month. Since mid 2017 I can't trust your numbers anymore and I still don't trust the corrected numbers until you change the whole methodology.
Even when 20 million people would do the survey at the same time (which does not happen) and the survey had 100 lines with 80 characters each it would only be 256 GB traffic in total, distributed to many servers - this is like downloading 5 AAA games for me, just in the other direction.
And I can promise you that less people will do the survey than can do it. So the traffic is not an argument.
"Historically, the survey used a client-side method to ensure that systems were counted only once per year, in order to provide an accurate picture of the entire Steam user population. It turns out, however, that many cyber cafes manage their hardware in a way that was causing their customers to be over counted. [...] It took us some time to root-cause the problem and deploy a fix, but we are confident that, as of April 2018, the Steam Hardware Survey is no longer over counting users."
The Fix doesn't make it accurate because how it must be done is assigning an ID to each steam installation or the system installation of the user.
But what if the user resets his Windows 10 to factory defaults and starts over installing Steam again because an broken Windows update killed his installation? -> Then the ID should also be counted as a new one.
What if a Linux user just changes his package sources to upgrade and never formats his computer just because it runs? He can only get the survey exactly 1 time per year.
But what if you run Steam in Wine for your 50 games with 50 wine versions? Don't the installations count as 50 different ones with 50 IDs?
What if you run a VM, with GPU passthrough and copy the virtual hard drive 20 times before you install Steam?
What if you get a brand new system image from the network each time you start your computer?
Some of these issues could be fixed perhaps but i am absolutely sure that you would only get accurate numbers when you only allow every Steam account (Edit: that has to exist for some time, e.g. one month and has to have a specific amount of value in the library) to do the survey on a client of their choice a single time per month. Since mid 2017 I can't trust your numbers anymore and I still don't trust the corrected numbers until you change the whole methodology.
Even when 20 million people would do the survey at the same time (which does not happen) and the survey had 100 lines with 80 characters each it would only be 256 GB traffic in total, distributed to many servers - this is like downloading 5 AAA games for me, just in the other direction.
And I can promise you that less people will do the survey than can do it. So the traffic is not an argument.