Not so simple
While I completely agree that some companies deliberately make things hard or impossible to repair, there are wrinkles (especially in electronics) that need to be considered.
1. Second source. When I design something I always try to use parts that are made by more than one manufacturer. For jelly bean parts, that is quite simple[1] but for more complex parts it is not. Many years ago, there were multiple sources of the 68xx families of microprocessors and microcontrollers. That is not the case with todays processors. There are huge amounts of microcontrollers and processors based on ARM cores but parts from different manufacturers are simply not interchangeable.
That means for any non-trivial design , there will be components that are only made by one manufacturer. Another area where this is the case is power regulators. If that company gets bought (lots of that has happened over the last few years) there is no guarantee I will be able to get that part in the future unless I do a rather expensive last time buy. I have actually had an EOL / EOB (end of life / end of buy) notice for well over 100 parts where the date on the notice was after the date of the last time buy so no-one could buy those parts any more.
The only parts where true interchangeability exists (because it was ever thus) is relatively simple logic devices. Anyone's 74LVC74 will operate the same as anyone else's, to a great extent (but be careful of timing parameters).
2. Prevention of reverse engineering[2]. The automotive sector is a case in point here. Companies making ECUs (which are now counted in dozens in some vehicles) don't want someone to make a cheap knock-off of their kit for various reasons so they take these steps:
Use parts that are pre-programmed by a factory and make it impossible to read the contents of program memory back (that capability has been around for well over 20 years).
Laser etch their own part number on it. The manufacturers of microcontrollers are only too happy to label a part with a vendor part number if 100s of thousands or more are being ordered.
3. Reliability. For many applications, using plug-in modules works great, but not so much in the high reliability sector. There is a rule of thumb in electronics that over 90% (some would say 95%) of faults are related to connectors. In addition to that, these things have an environment where soldered down memory devices are the only thing that makes sense. Reliable connectivity of fibre is a major headache in avionics as the 'standard' methods are laughably inadequate.
4. Component pin pitch and component density. Any decently equipped shop should have the ability to remove and replace components with very fine pin pitch (0.5mm is common as of perhaps 10 years ago) and BGA devices (standard DDR2/3 parts are in BGAs with 0.8mm pitch). Here I agree that repair should be possible but it requires a properly equipped shop. I personally have such equipment that I use for my own projects but it is not the sort of thing I would expect the average household to have.
5. The design and the designers. When the Apples and Samsungs of this world hear a clear demand for ever thinner gadgets, then something has to give and usually it is maintainability. As for the designers, very few of them have ever had to actually repair anything and as such they have no clue as to how to make something actually repairable.
There is more but you get the idea, perhaps.
[1]. Even the most mundane components often cannot be interchanged between manufacturers. Take the hugely popular 100nF ceramic MLCC surface mount capacitor (the go to choice for decoupling the power rails at ICs). They have interesting features such as DC bias and ageing issues. The variation between even the biggest players in the market makes it almost impossible to use different vendors except in the most undemanding of circuits. I certainly would never use different vendors if the loop stability of a circuit was even partially dependent on the characteristics of a surface mount ceramic capacitor. The same goes for resistors.
Even transistors with the same generic number are often not interchangeable from manufacturers. That bit me in the early 80s when a part from (what was) Motorola semiconductor products worked, but the part with the same generic number from RCA did not.
[2]. While there are certainly good reasons for this, it can be (and is) abused to prevent anyone else make or repair these things. Considering the regulatory and legal situation it may be more motivated by the desire to not be the target of a massive lawsuit.
Note that some semiconductor companies sell (and have been for at least 20 years) integrated devices to make changing a part without authentication impossible. One that stands out is Maxim Integrated.
Biting the hand that feeds IT
