How was it again with the proprietary systems?

From my point of view, the key issue is what "proprietary" actually means in the specific case: namely to what extent it refers to a network topology or a circuit diagram, a protocol or the concept of the joint or separate transport of power, signalling, and payload; or whether an actually standard system is artificially made proprietary by cooking its own stew at the operating system level and/or putting the whole thing on clay feet, so that the components have to be able to phone home like E.T., i.e. their brain is allocated in a server within the sphere of influence of the component manufacturer.

In my opinion, a good system uses freely available components whose function is based on the "basic operations" of switching / triggering impulses and has a control instance freely locatable in the cloud or on-site. From my point of view, this is a star wiring, a PLC in the house connection room and overall components that can make use of the classic resources of three- or five-core cables. I don’t want to need a SCART connector where a simple lamp socket would suffice.

Ceterum censeo: the three angels for Charlie should be Sabrina, Jill, and Kelly – and not Siri, Alexa, and Cortana :)

Apart from the fundamental discussion, it is still more economical to renew a new proprietary system every 5 years than to install KNX once for 25 years. KNX is simply disproportionately expensive.

That is way too generalized. For at least normal KNX installations, that is certainly not true.

KNX is even cheaper in direct and honest comparison than, for example, HomeMatic or Loxone. It only becomes disproportionately expensive if you end up with incompetent installers. The latter systems (and others as well) are usually programmed by oneself, so this approach must also be taken into account with KNX, and one should not, for example, counterpose an expensive professional configuration to the two.

renewing a new proprietary system every 5 years instead of installing KNX once for 25 years.

No, this approach is much more expensive overall. Because every manufacturer-specific system that is newly installed inevitably brings system and additional costs which lie outside of programming and the price of the components.

These would then have to be borne anew every 5 years.

KNX is simply disproportionately expensive.

But only if you compare apples to oranges. In an honest direct comparison, KNX is often even cheaper than many proprietary systems, not to say most, if you really compare everything exactly. Because even KNX is not exempt from development and prices constantly fall. Another advantage is also that you can still use the same wiring after 25 years and that exchanging and parameterizing the end devices is sufficient. Many system devices can still be used and essentially only need to be purchased once (power supply, interfaces, gateways, couplers etc.).

Well, no one has really been able to explain to me what the real, practical added value of KNX or other bus systems is compared to the simplest, cheap retrofit solutions like Shelly and co? Sure, I understand fail-safety, reliability, and durability. But technically, with the 'hack solutions,' I can realize almost all truly used scenarios for an absolute fraction of the price. I'm talking about 10-15K for KNX vs. a few hundred euros for the hack solution. What do you do with KNX that justifies these additional costs? Presence, lighting, shading, heating – I can hack all of that too.

One fiddles around, the other does it right away. ;) I especially like the variety of options with KNX, because with just one manufacturer there are always devices that I don’t like or that are less good in comparison. This way you can perfectly pick exactly what is optimal for the respective task. For our basic equipment alone, we already use eight different manufacturers. If I had to take everything from one, I would have to make compromises and would also pay more.