Month: May 2021

May 29 2021, 02:28

The second lecture is about eyes. Alexander Yakushev has a quite unusual diction and style of presenting the material – it’s somewhat emotional and with intonations, but essentially the lecture is very interesting.

However, it’s unclear where the lecturer got some of his information. For example, it is known that since the octopus’s eye evolved along a different evolutionary path, its color vision (as well as that of squid) is either absent or uses a different, still unclear principle compared to, say, mammals. It used to be believed that it sees with its skin (I have an unfinished book called The Soul of an Octopus by Montgomery lying in my phone – that’s exactly what is written there). Humans have three types of rods and cones, tuned to blue, green and red, whereas the octopus has only one type of photoreceptors, tuned to brightness (but it has many of them and its vision is sharper). How do we know it can distinguish colors? Because many species, like the chameleon, can replicate the pattern of their surroundings to blend in. So, Alexander Yakushev claims that their unusual pupil pulsates, and light reaches the retina at different angles. I never read about pulsation, but I did read that octopus uses an unusual eye shape (U or W shapes). Eventually, the eye focuses light waves of certain wavelengths on different parts of the retina. Color is determined through the brain processing different “snapshots” from the eye. Pulsation – is also an interesting theory, but usually people insert at least some reference to supporting works in their slides.

But overall, the lecture is short, and interesting with a set of facts about the nature of the eye.

https://www.youtube.com/watch?v=6Jsr9v6WH8c

May 28 2021, 23:30

Initially, I made thin plastic holders that were not meant to be removed from the car. The sun killed them. Then I replaced them with thick, but ugly wooden ones. They look eternal, but in reality, they are fragile. And today, I made an exact replica of the wooden ones, but so far from PLA plastic. It is biodegradable, does not like UV and direct sunlight, but unlike the first holders, it is still thick and, hopefully, stronger. Now the hooks are removable, so they can be taken off and not left under the scorching sun when not needed. On Monday, a new photopolymer printer will arrive, and I can try to print them from transparent green plastic.

Of course, all these experiments are not so much about making another phone holder. In general, the wooden one was enough for me. It’s just that in the process, I get to master new skills and abilities that would not have been acquired without the project. For example, to make this holder, it was necessary to figure out how to transfer the curvature profile to the computer (a cardboard that I scanned and outlined in Inkscape, saved in SVG), then transform the profile into a three-dimensional part (Autodesk Fusion 360), then refine the profile in Blender, figure out how to transfer it back from Blender to Fusion as a solid body (non-trivial), then conveniently cut the large part into sections connected by a screw and nut, virtually drill holes for them so that the nut is recessed, and eventually paint it all to protect the plastic from UV rays. Blender turned out to be a very interesting tool; I will continue to explore it further.

On Monday, they will deliver the photopolymer (stereolithography) printer, Creality LD-002R. So today I already have a phone holder model, I’ll try to print it on that as well. Unlike my existing printer, it prints with resin and the resolution is ten times higher (though it doesn’t matter in this case). Resin is not as strong as PLA plastic, but it is less vulnerable to UV, and there are no loads to worry about anyway.

May 28 2021, 10:27

A very funny story. But in general, I have worked in such companies in my time and generally understand how to interpret such situations. Well, from the perspective of a Software Architect.

I will tell you how a programmer with experience in optimizing processes in a large company.

The network of offices of any large company can be seen as a distributed system of servers with poor and unstable “hardware”. Each branch in this concept operates on its own program, which is generally a clone of one common program or its regional variant. Each such program at the local level works with its own set of data, and all this data is skewed and crooked. Accordingly, the program must take these errors in the data into account and not break down.

Any errors in the code of such a common program are painful because the entire network suffers. It takes years to polish it.

How well should such programs-processes cover all kinds of non-standard or rare situations? Obviously, you can’t foresee everything, so they wait until the number of incidents exceeds a certain threshold and then start to systematically solve it. If it doesn’t exceed, it’s cheaper to ignore it. Otherwise, the complexity of the processes will be such that errors will occur more frequently locally.

If the threshold is exceeded, it’s necessary to release and distribute a “patch”. This patch might fix one thing but break another. Therefore, the patch needs to be tested on a small network of branches, and then gradually extended to larger ones. It all takes a long time because it’s necessary to gather statistics and feedback, understand if it doesn’t break anything important, etc.

And in such a system, there’s always a huge list of features that need to be implemented and introduced. Essentially, it’s a very long backlog. Put yourself in the shoes of the owner of such a network – would you rather choose a feature that positively affects the revenue (plus 0.05%) or a feature that affects the satisfaction of 0.05% of customers?

Plus, very often inconveniences and issues simply don’t cross this threshold simply because a proper feedback system isn’t established. Customers don’t complain because “it has always been this way” and because “there’s nothing to compare to, they are the only ones in town”. Employees don’t complain because it works anyways, why add more work for themselves.

Besides, such a distributed system is constantly subjected to “attacks” and “hacking”. Not only by end-users but also by its operators – branch employees. This is when vulnerabilities begin to be used for personal gain. Often, it’s not even illegal. As in the situation with “tax optimization”, here, it involves using Russian ingenuity to exploit weaknesses in the processes for personal benefit. And it works like with the chairs from Ikea, which people take for a weekend picnic only to return them back to the store on Monday.

Any very large system is just full of such problems. They can only be solved by breaking the system into smaller pieces, which would compete with each other for the quality of service. This is how franchises are structured – take McDonald’s for example. It’s also important for more than one company to operate in the same field – competition fixes everything. This would fix “Post of Russia,” which is a monopolist.

May 26 2021, 10:47

A beautiful homemade phone holder was ultimately replaced by somewhat ugly wooden hooks. If you disregard their appearance, they turned out to be more convenient. Firstly, they are easily removable and installable. I don’t always need them, and there’s no unnecessary bulk. Secondly, they fit much tighter, nothing wobbles. Thirdly, I finally got around to using the jigsaw and paint, which seems more important than the result.

The previous holder, printed on a 3D printer, did not survive our sun. Ultraviolet combined with hot rays deforms PLA plastic, and my printer can’t handle PETG. There’s little hope for photopolymer printing since the resin also doesn’t particularly like direct sunlight. So, the temporary wooden alternative might easily become permanent 😉