Matrix, AI vs Humans, Software Projects
Matrix - Did I say Matrix is broken? Media shares in offline mode do not work, as well as using "share" from other applications with Element X on Android (Matrix client for Android), makes it almost certainly to crash. These are again simple things which work for everyone else. And if you mention these shortcomings in a Matrix community, you'll get attacked by them. - Well, it is what it is. The brown emoji here. - Same applies to the login flow, the device you're confirming the login from, shows that are you now logging in with new session JFCGEJQS and guess what the new device which is waiting for the login confirmation, doesn't show ANY REFERENCE. So sad, and that's really BAD UX. No matter what you say, "yes it works", but it's confusing and very error prone and quite brown stuff.
Matrix - Still way too many serious issues for regular users, can't recommend (list list too long). SimpleX: Solid for techies once a couple of things are fixed (group efficiency, key and contact backup, alternate backup routing). Otherwise, have to admit WhatsApp & Signal just work reliably, even if those got well known shortcomings (centralized platform, key management). - Also Matrix encryption / identity backup key feature was changed in a way, that the keys users had stored didn't work anymore. Those are major flaws causing potential data loss.
AI vs Humans - Actually AI ranks much better against humans, when we take reality into the count. - Many friends claim that AI gives incorrect answers. But when you technically and truly analyze the question and answer, it's not incorrect. They just didn't specify clearly enough what they wanted to be answered to. They THOUGHT about something, but they didn't ASK IT. - Similarly, even when asking something, many people first create bad question, and the person answering the the question doesn't pay enough attention to the question and then just answers something. - This is the unfortunate reality. - We can distill it down to: "AI's practical efficacy is amplified by prevalent human deficits in both cognitive diligence and communicative precision." - Funny part is that I had to have several iterations with AI to get my request straight for getting that statement out. So, that absolutely underlines what it says.
And a few longer format posts follow.
AI vs. Human Performance: A Testing Proposal
If AI responses are poor, perhaps we should conduct a real-world test: Gemini 3 Flash vs. Humans.
Let's take questions from last year's matriculation examinations. We'll promise € 1 as a reward for successfully completing a task. We could go to a Lidl parking lot on a Saturday morning, stop people, and offer them money (that one euro) to try. What do you think the success rate would be, even among those who agree to attempt the questions?
Software Development Challenges and AI Testing Proposal
The Human Factor in Software Projects
They claim, AI is worse than humans. First question is of course, are you comparing tasks performed with the same resources or completely different resources?
My personal experience with software development and project management (spanning several decades) is that it actually specifically highlights people's inadequacies and unsuitability for tasks.
The client's desires are (almost always) unclear; they don't even know what they want
Specifications are inadequate (always), often practically non-existent
The entire project can be based on incorrect premises (fortunately not as often)
Things are endlessly tweaked and iterated (agile methodologies)
Requested features are poorly implemented, even when specifications are technically perfect (which they never are)
Testing and feedback are insufficient; nobody cares (they just want to mark it as complete, even if it doesn't actually work)
And this is supposed to be a good result? In theory, things could go differently, but usually they don't. In my opinion, this particularly highlights human inadequacy, which boils down to a few factors:
Lack of motivation
Lack of understanding
Poor communication
I believe all the aforementioned issues are essentially basic human characteristics.
2025-12-21