Thanks for the comments, Roman. Each of your many points might require a separate discussion. Nevertheless, I will try to briefly respond to each of them.
You write “Scientists and programmers understand “model” differently”. Could you tell the difference?
“Descriptive/performative”. I agree with you. It does not change the point of the article.
“if we take ‘scientific’ model, it does not have anything ‘performative’”. I am not sure I understand your point. My point is precisely that modeling, either in science or in programming exactly lacks performativity. The reason is that performativity is a linguistic thing. It is in the blind spot of modeling.
“No amount of linguistics and human ‘common sense’ can replace an algorithm”. It is not about replacing algorithms. It is about complementing algorithms with sound semantic/pragmatic analysis.
“I can’t agree modelling has no place in software development”. I never said one should not use models. “Biased” does not mean “wrong”, it means “excessively reliant upon”.
Ontologies: If you want to talk about them, why don’t you pick a simple ontology and demonstrate how one can deduce usage from it. The point is that the ontology/vocabulary usage or deployment is outside of any semantics — it’s a pragmatic step, which is straightforward in our use of language (natural or artificial), but tedious and often intractable, if you want to make hierarchies of definitions your point of departure.
Your example of chess is excellent (volumes are written on that in the context of rule following). Assume you explain chess to a person (e.g. a child) who does not know what “game” is: he does not understand why one should take turns, why one even might want to win, etc. You will find yourself building endlessly extensive ontology for such a person (read Wittgenstein or Kripke on rule following, that’s exactly about that). On the other hand, you pretty easily could teach someone about games, winning, rules, etc on examples, exercises, etc. As another example, notice how over the past two decades software tutorials almost completely have moved away from definitions to learning through examples.
Ontologies not always are the best tool for a job. They are not a universal knowledge or meaning tool, simply because they don’t account for certain properties of the language.
“regardless of whether we use OOP or FP or some other P, we deal with ontologies. This is basic mode, so fundamental, that it came originally from philosophy (actually, like many other concepts in linguistics).” Well, my main point is that modeling (and ontologies) are essentially limited. Seeing them as universal is exactly a flavour of ‘modeling bias’. Ontologies have their strengths, but they are not universal. My couple of articles on Medium are trying to show exactly that.
