I can sum up the linked article in a few succint words: "I paid $2,000 for hardware, and no one would hold my hand and tell me how to connect it together. I had to learn how to do it."
His beefs are about running cables and mounting components, and includes the following paean to Apple:
This is why people buy from Apple. It designs everything from the trackpad to the box the computer comes in, which unfolds neatly to reveal everything you need. Apple reduces friction to the point where even my mom could upgrade the RAM on her iMac, and it can do this because it controls everything that goes in that box....which is why people are still building PCs to get maximum performance for gaming.
That's why I recommend Apple products to people who aren't tech savvy. They just work. When I'm pushing a water cooler down on the CPU while twisting its radiator into place and screwing it into place at the same time, it becomes clear that PCs don't just work.
(Admittedly this image is from 2009. Still, most Macs, you still can't put in an aftermarket video card.)
It really isn't that difficult. If you've never put a PC together before, yes, you're going to have to learn how to do it. But all the really hard stuff has been done for you already, and all you need to do is to put the components together the right way.
(If you're going to insist on water cooling? That's on you, pal. That adds a big layer of complexity, one which is not strictly necessary. It's only really necessary if you're intending to overclock; and if you're not proficient at building a PC you've got no business overclocking.)
Building a computer in 2016 is incredibly easy. Let's have a look at the steps required to build a computer in 1991:
1) Install processor in motherboard. Generally this has not changed, and in fact was a little easier back then.
2) Install memory. For the most part, by 1991 computers were no longer using discrete ICs but instead had moved to SIMMs, the 30-pin variety. This was also fairly easy and hasn't changed in difficulty since then. If you had a computer which used SIPP memory (30 pin) you had to line up 30 pins to the socket on the mobo, but still not too hard.
3) Install standoffs in motherboard and case. This also hasn't changed. You needed to make sure that the screw-down standoffs weren't shorting something important on the mobo, though, and not every mobo used the same pattern of plastic and screw-type standoffs. Took a bit of finagling, but not hard, and still something you might have to do today.
4) Configure speed display. Most cases had 7-segment LED displays showing you the clock speed of the motherboard in both normal and "turbo" mode. Yes, many machines had switchable clock speeds, because some software wouldn't work right at higher speeds (like games). If you wanted your LED display to reflect those speeds (in megahertz, MHz) you'd spend ten minutes deciphering a crinkly sheet of jumper settings.
5) Install motherboard. Also about the same as today, you'd plug in the power connector and then get all the fiddly little case wires in place, like for the hard drive light, power light, reset and turbo switches, and so on. If there were configuration switches or jumpers on the motherboard now was the time to set them, too.
But now things start to get difficult.
6) Install interface cards. Motherboards came with a keyboard connector. If you were lucky, the board might also have on-board serial and parallel ports, but it was not very common for clone PCs. Even then, those ports had to be configured in hardware exactly the same way the ports on the cards did. And so what you'd do is you'd decide where in memory, and what hardware interrupts, LPT1 and COM1 and COM2 would occupy. There were standardized settings for them but the cards usually had to be double-checked for the correct settings, and if you tried to put two ports on the same IRQ you'd have a mess on your hands.
All the settings were done with jumpers (if it was a cheap card) or switches (if it was expensive) but the result was the same: manual configuration of everything.
So--assured that the cards were all addressed correctly--your next step was to install the drives, floppies and hard drive. No CD-ROM. No IDE interface, either.
You typically put in both a 5.25" and 3.5" floppy drive. It was up to the customer which would be drive A. When installing the cabling, you had to use an adapter to get power to the 3.5" drive because the standard Molex connector was too big for one. You had to make sure that drive A was on one side of the twist in the cable, and drive B was on the other. Frequently you needed an adapter to plug the floppy cable into the 3.5" drive, too, because it used a header instead of a card-edge connector. The 3.5" required a mounting kit, too, because all the drive bays were 5.25". (At this time, you still ran across floppy drives which required terminating resistors. See the discussion for hard drives on what that was like.)
The hard drive was its own set of problems. As I said, "no IDE interface" (unless this was an expensive machine) so you were running MFM or RLL, and had a separate controller plugged into a card slot. For a single-drive system, two cables (one for commands, one for data) ran from this card to the hard drive. If you had a second hard drive in the system, not only did you have a second data cable running from the controller to the second drive, but you also had to make sure that the hard drives had the correct terminating resistors. The typical hard drive had three 8-pin SIP terminating resistors, and you had to know when to remove them and when to leave them. The drive in the middle of the command cable would be drive 1 and the one on the end would be 0--again, there was a half-twist in some wires between the two connectors--and 1's terminating resistors had to be removed or operation would be "unpredictable".
7) First power-on. So, you finally got all the wires in and switches and terminating resistors configured. The motherboard was set up to recognize the hard drive type (usually with DIP switches) and everything was good to go. With everything was finally configured and plugged in, then it was time to turn on the power.
If you were lucky--or good at building machines, as I was, eventually--the thing would boot up and present you with a power-on self-test (POST) screen, and after it had finished counting RAM and running its other checks the thing would try to boot. DOS in drive A would get you to a A:\ prompt. If it did not go to an A prompt--or, worse, if it just sat there and beeped at you--you got to dig into the hardware again to find what you missed. In these cases you'd find a SIMM not installed right, or that you'd plugged in a 3.5" floppy backward, or-or-or. It rarely was because a hardware part was bad; it usually was because you'd made a mistake.
8) Prepare hard disk for use. Once you got the thing to boot up, you got to format the hard drive. Step one would be to put in the low-level format utility for your brand of hard drive and run the program. You needed to know what the physical parameters of the drive were; and to make things even more entertaining you could only (at that time) make a 32 MB partition, so if you had a Seagate ST-251 hard drive you had the option of splitting it into two 20 MB partitions or one 32 and one 8 MB partition. Furthermore, the drive type table in the motherboard's BIOS only had a handful of drive types available. The ST-251 was usually type 40; if you weren't using a drive that was in that table, God help you, because your job was going to be much harder.
But you had to low-level format the drive first. The program would take the physical parameters of the hard drive (cylinders, heads, sectors per track) and then lay that information down on the hard drive platters which--until just now--were utterly blank. Most drives had a label on them showing which sectors had been marked as defective at the factory; you had to input this information into the low-level formatting program. Then you went off and had a snack while the drive low-leveled.
Drive not in the drive table? You get to use the manufacturer's program to trick the computer into thinking it's a standard type. It added a layer of complexity, and if that little bit of code ever got corrupted, the entire drive became 100% inaccessible. It was a lousy way to do things and I strongly discouraged it whenever possible.
Low-level done, you could do the high-level, which took less time but still wasn't fast. After that was done, you could then reboot the computer and mirabile visu be presented with the C:> prompt.
9) Install OS. Installing the OS meant copying DOS from floppies to the C:\DOS directory. You'd want to set up an autoexec file, too--AUTOEXEC.BAT, which would run whenever the computer was rebooted--something like this:
@echo off...which told the machine to look in C:\DOS whenever a command was typed. You could add to the path, so (for example) if you wanted to be able to run EVERCRAK.EXE without first typing CD EVERCRAK you could modify your path statement to say something like "path=C:\dos,c:\evercrak". Anywhere you were in the file system, if you wanted to run a command you need only type that command and the computer would dutifully check every entry in the path system variable for that command, and it would run the first example of that command it found. (Let's say you type GOOP to run that program. GOOP 1.0 is in C:\DOS but GOOP 2.0 is in c:\GOOP. If DOS is ahead of GOOP in the path variable you'll get GOOP 1.0 every time, unless you first go to the GOOP directory.)
You also would typically write a CONFIG.SYS file:
files=40You didn't need much.
At this point, then, the computer would be ready for software installation, and you could do what you needed for that. If for any reason you had to use nonstandard parameters for your I/O ports (LPT and COM) you have to tell the software what to use.
Before Windows 95, it got worse, because of the memory management tools for DOS 6 and the addition of multimedia devices; you had to load drivers and put all kinds of things in your configuration files--but fortunately the installation utilities for those devices would usually handle all that.
Compare that, then, with assembling a computer in 2016. All the hardware configuration stuff is automatic. For one thing, nearly all the ports you need are integrated onto the motherboard; you no longer need anything for a standard office PC but a motherboard, a hard drive, and an optical drive. Once you've got the hardware installed in the case, the cables are all plug-and-play (they can only go in one way and it's one signal and one power cable per drive thanks to SATA) and the computer sets all the interrupts and memory locations automatically.
For a computer that needs better video, you plug a video card in. No jumpers, no configuration; the computer knows what to do with it. You can assemble a computer from basic parts in half an hour if you've done it before. Once you've got it booting up you throw an OS disk into the optical drive and then spend an hour clicking options. The hard drive is preformatted at the factory, so you need only run the high-level format (which is good because low-level formatting a terabyte drive would take days). (Slight exaggeration.)
So: are gaming PCs hard to build? Not even remotely. Not compared to the typical office PC circa 1991.