The biggest problem with the new icons is not lack of beauty but that the overly stylistic design has made them more difficult to visually parse.

The purpose of icons is to make recognition of objects on the screen easier. The use of three dimensions, contrasting edges, shading and shadows are significant visual aids - and those are the things that these new icons lack the most. It takes more than Photoshop skills to earn the title of UX Designer.

It is not about having ads, but the nature of the ads. Ads that blink, flash, have lots of moving objects .. and worse of all: play sound.

There is a reason why some browsers have had the feature to disable GIF animation for many years. Until recently, the majority of animated ads were made in Adobe Flash, which you could have configured as click-to-play.
With HTML5 and the most popular browsers, there is no click-to-play.

Different schools of touch-typing assign the digit 6 key to different hands. This goes back a long way.
We talked about this at a keyboard-oriented discussion forum recently and one user had found two different touch-typing manuals in English from 1889 and 1893 that were different about this key.

The original Scholes and Glidden QWERTY layout used the letter I as the digit 1. The numeric row started with the digit 2, so the whole row was shifted one step to the left compared to modern keyboards.
So originally, the number 6 was truly on the left side but moved to the right when the numeric row got a proper 1-key.

I once did a survey of split ergonomic keyboards. The most popular series: the Microsoft Natural Keyboards, has it on the left.
Out of 24 keyboard models, 16 had the number-6 key on the left side, 7 had it on the right side and one even had it on both.

Most people don't use the number pad on a full-size keyboard.
The function keys are largely superfluous if you are using anything but MS Windows.
Most people press the Space bar in exact the position where it is located on this keyboard.
Using the mouse too far to the right, past arrow and numpad contributes to shoulder problems. ... and I have met lots of programmers who use emacs or vi who don't need the cursor keys or the nav cluster.

This form factor is actually quite popular among professionals who do a lot of typing. Just look at the Happy Hacking Keyboard, Poker and Ducky Mini, to name a few. They are small premium keyboards, and they sell.

The big feature of this keyboard that separates it from others is that it splits into two. I can tell you from having tried many ergonomic keyboards that just separating the hands a bit is the most significant feature that an ergonomic keyboard could have. Besides that, you can position and tilt the halves the way you want.
You could view this keyboard as a more portable alternative to the Microsoft Natural Keyboard if you will.

3D-printing is overkill.

When I buy a new mouse, I remove the mousewheel. Then I build a middle mouse button that fits in its place. I cut pieces of styrene and put them together with plastic cement ("model glue").

I use the middle mouse button mostly to open new tabs/windows in web browsing where invoking the scroll wheel inadvertently causes the page to scroll away from where I'm clicking. That is very annoying.

Most mice I have used use the same type of microswitch for the "middle button" as for the left and right buttons, but in most mice the buttons are levers with a rod on the microswitch.
These levers are usually the same on the left and right buttons, but very different on the middle press so the sensation is very different.

But most of Stargate SG-1 was aired in the 00-ies.

Yes, but instead of having a status register, you compare each item in one vector with each vector in another and get the results as a vector of booleans.
Then execute a SIMD instruction, where each component scalar operation is conditional according to each corresponding boolean.

Or, you could convert that vector of booleans into something else. For instance, you could count the number of leading 1's in the vector and store into a scalar, which would allow you implement operations such as strlen() or strcmp() with vectors.
(It is a bit like programming in APL, if you have tried it)

These types of operations have hitherto mostly been done by DSPs.
An architecture for general-purpose computing under development that would do this well is The Mill. Mind you, it is very interesting in other ways. There is a lot of stuff about it on the web site, and good talks about various features on Youtube.

I agree with what you say, but I would like to add curved to that list of things to look for now that curved screens are coming.
While curved TVs are nothing more than a marketing ploy, having ultra-wide computer monitors be curved makes a lot of sense. I've never seen anyone use a two-monitor setup with both in the same plane - always at an angle to each other.

However, I have heard said that curved computer screens would be worse for graphics design/editing work. I don't know if that refers to them not being flat, or if accurate-enough colour reproduction isn't available in curved displays yet.

I want to see functional and well-designed stuff. Things that contain the capabilities that people like, those capabilities done right to modern specs and the absence of cruft.
Apple under Steve Jobs did manage to follow this ideal, except that they liked to lock people into their system and upgrade often. Lenovo has also followed this ideal, mostly.. except for their consumer space.

For instance, I am not interested in any Windows tablet without a stylus.
No curved TVs.
No laptop that requires you to open the lid to turn it on while it is "docked".
No "smart watches" that need to be recharged every eight hours, or every four hours after two years because the non-replaceable battery has degraded.

Please!

Indeed something along that line is what I think the Internet protocol needs. While IP is freely packet-switched and may appear stateless when you glance in the specs, TCP/IP routers and hosts are actually session-based internally and the number of concurrent sessions is limited.

It is not only intentionally malicious code that can cause denial of service: legitimate programs that are merely badly designed can also do it.
Then it is not the network and the other services running over it that should be punished by being throttled, but only the individual node or badly behaving program.

Also, what we don't need is something that could restrict innovation in new protocols over TCP/IP. If the Internet infrastructure would allow not much more than only email and HTTP/HTTPS (which some ISPs are doing in some countries), then attackers are just going to find another attack vector .. on top of a TCP/IP that permits it.

The point is that you are supposed to hold it for a whole year.

My new year's resolution is 210 PPI.