The original name of Genius was RapExegesis, which only confirms the main point of the article: that Genius' founders were mostly pedantic Yale grads. /s
Gettiers in software engineering (2019)
151–160 of 222 posts
Re: Gettiers in software engineering (2019)
#152Earlier quoted context omitted.
Bayes theorem isnt a reasonable approximation, because it isnt answering the question -- it describes what you do when you have the answer. With bayes, you're computing P(Model|Evidence) -- but this doesnt explain where Model comes from or why Evidence is relevant to model. If you compute P(AllPossibleModels|AllSensoryInput) you end up never learning anything. What's happening with animals is that we have a certain,…
I'm not sure I'm understanding your stance fully, so please forgive any poor interpretation. > certain, deterministic, non-bayesian primitive model of our bodies What makes you certain the model of our body is non-Bayesian? Does this imply we have an innate model of our body and how it operates in space? I could be convinced that babies don't inherently have a model of their bodies (or that they control their bodies)…
Because bayesian conditioning doesn't resolve the direction of causation, and gives no way of getting 'certain' data which is an assumption of the method (, as well as assuming relevance).
In bayesian approaches it's assumed we have some implicity metatheory which gives us how the data relates to the model, so really all bayesian formulae should have an implicit 'Theory' condition which provides, eg., the actual probability value:
P(Model|Evidence, Theory(Model, Evidence))
The problem is there's no way of building such a theory using bayesianism, it ends in a kind of obvious regress: P(P(P(M|E, T1)|T2)|T3,...)
What theory provides the meaning of 'the most basic data'? ie., how it relates to the model? (and eg., how we compute such a probability).
The answer to all these problems is: the body. The body resolves the direction of causation, it also bootstraps reasoning.
In order to compute P(ShapeOfCup|GraspOnCup, Theory(Grasp, Shape)), I first (in early childhood) build such a theory by computing P(ShapeSensaton|do(GraspMovemnt), BasicTheory(BasicMotorActions, BasicSensations).
Were 'do' is non-bayesian conditioning, ie., it denotes the probability distribution which arises specifically from causal intervention. And "BasicTheory" has to be in-built.
In philosophical terms, the "BasicTheory" is something like Kant's synthetic a priori -- though there's many views on it. Most philosophers have realised, long before contemporary stats, that you cannot resolve the under-determination of theory by evidence without a prior theory.
Re: Gettiers in software engineering (2019)
#153Earlier quoted context omitted.
Bayes theorem isnt a reasonable approximation, because it isnt answering the question -- it describes what you do when you have the answer. With bayes, you're computing P(Model|Evidence) -- but this doesnt explain where Model comes from or why Evidence is relevant to model. If you compute P(AllPossibleModels|AllSensoryInput) you end up never learning anything. What's happening with animals is that we have a certain,…
> If you compute P(AllPossibleModels|AllSensoryInput) you end up never learning anything. ...whoa. That makes complete sense. So you're saying that there must be some form of meta-rationality that gives cues to our attempts at Bayesian reasoning, directing those attempts how to make selections from each set (the set of all possible models and the set of all sensory inputs) in order to produce results that constitute…
In order to think, we move.
Re: Gettiers in software engineering (2019)
#154An old timer I worked with during my first internship called these kinds of issues "the law of coincidental failures" and I took it to heart. I try a lot of obvious things when debugging to ascertain the truth. Like, does undoing my entire change fix the bug?
I wonder if there is a law of coincidental succeeses too. (if you're an old timer, you might call this some sort of Mr. Magoo law, or maybe "it seems to work for me")
This is the root of 'pigeon religions'. Someone sees a string of events and infers a causal link between them and an outcome. Confirmation bias kicks in and they notice when this string of events occurs again, which is made more likely by the fact that the events in the string are largely the person's own actions which they believe the events will produce the desired outcome. They tell their friends and soon a whole group of people believe that doing those things is necessary to produce that outcome.
That's how you get things like equipment operators insisting that you have to adjust the seat before the boot will open.
Re: Gettiers in software engineering (2019)
#155One way out is to just admit that "justified true belief" is not a satisfactory definition of knowledge. I know that's not really the point of this article, I'm just saying.
Re: Gettiers in software engineering (2019)
#156Earlier quoted context omitted.
I often wonder if LLMs would have made Wittgenstein cry...
I don't know, but I reckon he would have been unimpressed.
I would argue the opposite; LLM technology is the first viable means to computing "language games" as such, and quite in line with the late W. theory.
Re: Gettiers in software engineering (2019)
#157Earlier quoted context omitted.
edit : Then there's the whole "what is a cow" thing. Like if you you stuffed a cow carcass with a robot and no one could tell the difference, would that still be a cow? Or what if you came across a horrifying cow-horse hybrid, what would you call that? Or if the cow in question had a unique mutation possessed by no other cow - does it still fit the cow archetype? For example, what if the cow couldn't produce milk? Or…
A robot in a cow carcass is not a cow, it's a "robot in a cow carcass". Someone might believe it's a cow because they lack crucial information but that's on them, doesn't change the fact. A cow-horse hybris is not a cow, it's a cow-horse hybrid. A cow with a genetic mutation is a cow with a genetic mutation. A cow created in a lab, perhaps even grown 100% by artificial means in-vitro is of course still a cow since it…
I have this pet theory that Philosophy is kind of the Alternative Medicine of intellectual pursuits. In the same way that Alternative Medicine is doomed to consist of stuff that doesn't work (because anything proven to work becomes "Medicine"), Philosophy is made entirely of ideas that can't be validated through observation (because then they'd be Science), and also can't be rigorously formalized (because then they'd be Math).
So for any given claim in Philosophy, if you could find a way to either (a) compare it to the world or (b) state it in unambiguous symbolic terms, then we'd stop calling it Philosophy. As a result it seems like the discipline is doomed to consist of unresolvable debates where none of the participants even define their terms quite the same way.
Crazy idea, or no?
Re: Gettiers in software engineering (2019)
#158Would a merge instead of a rebase have made it easier to find the bug? (Serious question)
In this case it would not, you can see the commits between tag A and tag B either way. He simply didn't bother checking, either that there was no bug after rebasing or that the commits he'd rebased onto hadn't been released yet.
Re: Gettiers in software engineering (2019)
#159Question I like to ask my colleagues. Suppose you have a program that passes all the tests. Suppose also that in that program there is a piece of code performs an operation incorrectly. The result of that operation is used in another part of the code that also performs an operation incorrectly, but in such a way that the tested outcome is correct. Does the code have 0 defects, 1 defect, or 2 defects?
I can understand 0 and 2, but what's 1? The "I don't know how to count" answer?
The second part could instead be more like "detects an edge case and handles it, accidentally working around the first defect".
Re: Gettiers in software engineering (2019)
#160I'm not sure I see the big deal. Justification is on a scale of 0 to 1, and at 1 you are onmiscient. We live in a complicated world; no one has time to be God so you just accept your 0.5 JTB and move on. Or for the belief part, well, "it's not a lie if you believe it". And as for the true bit, let's assume that there really is a cow, but before you can call someone over to verify your JTB, an alien abducts the cow an…
You're view is more inline with the philosophy of science which holds nothing an ever be justified. https://www.wikiwand.com/en/articles/Karl_Popper read The problem of induction and demarcation: https://www.wikiwand.com/en/articles/Falsifiability Basically to some it all up because we aren't "omniscient" nothing can in actuallity ever be known.
Well, nothing can be truly known on _inductive_ basis but we surely can know things on _deductive_ basis