Windows 10, 32 GB of RAM, and the RTX 4060 all surge to the top spot in the Steam Hardware Survey, seemingly due to a huge influx of Chinese-language user data

It’s a new month, so you know what that means, right? Yes, it’s time to dive into Valve’s latest figures in its Steam Hardware Survey and you don’t need to be an expert in PC gaming to see some unusual trends. Top operating system? Windows 10. Most common amount of system RAM? 32 GB. One possible answer for all of this might be found in another statistic, with the most common user language being Simplified Chinese—up 21% compared to last month.

Every month, Valve randomly pings a sample of its Steam userbase, prompting the app to ask if you’d be willing to let it scan your PC for its hardware configuration and upload the data. We then see the statistics from this sample in the Steam Hardware Survey, and the February results are certainly a little different from previous ones.

For one, Windows 10 now holds the top OS position, with an increase of almost 11% from January, despite Windows 11 previously being the most common. Likewise, the amount of system RAM has increased, with 32 GB apparently being the most popular, with a rise of just under 14%. Nvidia’s RTX 4060 has displaced the RTX 3060 as being the top GPU, nudging ahead thanks to a 4% gain.

Has everyone suddenly just switched back to Windows 10 and bought a pile more RAM for their gaming rigs? Well, possibly, but the most likely explanation for these rather large changes in operating system and memory configurations can be found by looking at the most popular user language in the survey. Typically, it flip flops between English and Simplified Chinese on a monthly basis, but for February, the latter shot up 20.88% to garner 50.06% of the survey.

The RTX 4060 gains are less of a surprise, though, as there was only a small gap between it and the RTX 3060 back in last November. However, a 4% rise compared to February is quite a leap in the world of GPUs, as most changes are very small.

Now, this doesn’t mean that there’s anything screwy going on, far from it. China’s PC gaming market is huge, so if Valve is taking a fully randomised sample of its userbase, then there’s a good chance that the Chinese market will form a significant portion of the survey. However, in the case of February, the 21% increase suggests that this time round, the sample is statistically skewed. Still relevant but it does mean that one needs to view the changes in OS, RAM, GPU, etc with a degree of caution.

One change that Valve could implement to make the hardware survey results really useful is to show the same statistics collated for each major geographical market, e.g., the US, EU, SEA, etc. That way, one could see, at a glance, just how much the results are skewed toward a particular region. At the risk of repeating myself, skewed information isn’t bad information, as long as one is fully aware of the nature and degree of the skew.

Another change that I’d like to see, is a setting added to Steam whereby your PC is automatically scanned for hardware once per month. I hasten to add that this should not be enabled by default and require several ‘approval’ steps to activate, but it would help to get around a weakness in the surveying method.

Valve may possibly ping the same number of users for each sample and it might even be the case that the system keeps pinging away until it’s accumulated a set number of responses. Either way, though, the survey results are only valid for users who have responded.

I’ve contacted Valve to see what its thoughts are on the above and it’ll be interesting to hear what it has to say on the matter. For now, it’s best to not read too much into the Steam Hardware Survey’s results. It may well be true that 32 GB is the most common amount of system RAM in Steam users’ PCs, but without more information, one can’t say that with any degree of confidence.

Last month’s survey results went live with a few greebles, only to be rapidly adjusted to show the correct figures, but I’m confident that these ones are right, just hugely skewed by a specific market. The only problem is that I can’t be sure about that without more information, which doesn’t help the survey’s usefulness.

What did Mark Twain write? ‘Lies, damned lies, and statistics.’