In 2015 I had 2 4k60 monitors, and as long as I setup screen composition in the nvidia settings, everything was as smooth as silk and as sharp as a magazine in Linux, and still is.

In 2015 I was in CLion day in and day out and my gpu was only a laptop gpu with 2gb of vram, so it definitely maxed out my gpu then, but still was as smooth as butter. I had to worry about my computer heating up at the time.

Today I'm on a desktop with a 980. I'm sure it's inefficient, but doing anything in the desktop, like watching youtube in 4k60fps uses about 15% of the gpu according to nvidia-smi. With all my apps running, when I'm not training neural networks, my desktop + firefox takes between 1gb and 1.5gb of vram.

That is unfortunate.

I run two HP Z32 4k monitors side by side in portrait mode. Running Debian Linux Buster. Connected to NUC8i7hvk. LXDE, sometimes KDE. Text is clear. Moving windows around is smooth as silk.

27" would be too small. 30" is about the right size for the resolution.

Plus I have five or six virtual desktops to task switch between various development projects.

I run Linux as well; Went from 1920x1200 (the fabulous 16:10 ratio) to 2560x1440 (16:9) and would never go back down. I'm using a triple display setup with 1440p, and they operate at 75Hz no problem (no input/output lag). Coding no problem, multi-task no problem, everything is just peachy. I suspect the jump up to 4k is just symptomatic of crossing the boundary of acceptable image scale. Apparently 1440 approaches the boundary, but 4K is well beyond. My GPU is not very fabulous, and in fact is the limited GPU available with an Intel NUC (Hades Canyon), but the connector to the display is using DisplayPort, and that I believe is the point the OP is trying to make with the Opinion Piece. I find this conclusion troubling because DP is a few thing: a cable standard (shielding, twists of pairs, etc), a differential signaling protocol, and the features layered on top. A lot of people conflate one aspect with the other, which quickly becomes problematic. For example, the DP protocol was incorporated into thunderbolt 3, so the high level stuff mostly, protocol, etc. The thunderbolt 3 cabling sufficiently meets the standard for the cable standard parts (shielding, isolation, etc). I guess where i'm going with this is that HDMI is slightly more problematic here, especially in terms of the matrices of cable standards versus protocol standards, and the consumer buying the cable or understanding what protocol they need, etc... HDMI made the mistake of introducing a kind of "high speed cable" in the HDMI 1.x protocol era, rather than simply jumping to HDMI 2.x, that is to say NOT aligning major spec jumps to to physical cable requirements, but instead to protocol features. It's probably not a fair comparison with DP and HDMI, since DP sorta didn't have the issue, but it becomes apparent that the future generations of DP cable bring physical cable requirements with the next major version bump, protocol topics are there too but we can get to those later. So for example, DP being in Thunderbolt 3, and now Thunderbolt 4 being drafted, and and.. the combination of Thunderbolt 4 with USB 4... which is capable of transporting HDMI 2.x spec protocols. Ugh... So it seems the physical cable specs of video cables is coming to an end, except for high-end video products (e.g. 8K and beyond). In the general purpose (4K and less) sense there will be one cable for both HDMI and DP (where DP is the default protocol). So all that said, I don't really see HDMI the protocol being a problem, but certainly HDMI cables and interfaces are a problem, perhaps not so big a problem considering most modern cables meet or exceed the requirements. But there are outliers, and there is the problem.