uniqueuid

10 ms·

61.
by uniqueuid 1y ago
I wonder what kind of edge cases you deal with when blas is your bottleneck in R. Stan code aside, I’ve seen few problems that are neither instant (i.e. sub hours) nor impossible (I.e years of compute).
62.
by uniqueuid 1y ago
The causal links in this analysis are very heterogeneous. Some causes for accelerated aging seem relatively direct and plausible with causal models that have supporting literature, i.e. air quality. On the societal level, it's much mor
63.
by uniqueuid 1y ago
This paper builds on a series of pathways towards harm. Those are plausible in principle, but we still have frustratingly little evidence of the magnitude of such harms in the field. To solve the question of whether or not these harms can&#
64.
by uniqueuid 1y ago
The counterpoint here is that in practice, humility is only found in the best of frequentists, whereas the rest succumb to hubris (i.e. the cult of irrelevant precisions).
65.
by uniqueuid 1y ago
Popper requires you to posit null hypotheses to falsify (although there are different schools of thought on what exactly you need to specify in advance [1]). Bayesianism requires you to assume / formalize your prior belief about the su
66.
by uniqueuid 1y ago
My experience is that it is always important to criticize free speech absolutism, especially when people behave as if it were an atemporal concept. In reality, most of the world for most of the time has had various compromises between pro
67.
by uniqueuid 1y ago
The word 'natural' is doing titanic work in your sentences here ... :)
68.
by uniqueuid 1y ago
Wow, that must be quite expensive! You said the files alone are a few PB. So at least 2PB / 8 servers ~= 250TB per server, which would probably put each server at > 20k $ (unless you’re putting it together with duct tape and scraps,
69.
by uniqueuid 1y ago
That sums it up exceptionally well.
70.
by uniqueuid 1y ago
Well I'd say that prompt engineering is still exactly this?
71.
by uniqueuid 1y ago
While I get your point, it doesn't carry too much weight, because you can (and we often read this) claim the opposite: Linear regression, for all its faults, forces you to be very selective about parameters that you believe to be meani
72.
by uniqueuid 1y ago
Cats also have timeouts set for balls of fluff. They usually get bored at some point and either go away or attack you :)
73.
by uniqueuid 1y ago
Practically all standard libraries have timeouts set for such requests, unless you are explicitly offering streams which they would skip.
74.
by uniqueuid 1y ago
Well they do say they calculate propensity scores, but the whole language of the blog post is very hand-wavy. Warning over and over again that something is not causal proof, and then dramaticizing it as if it were proof is just unprofession
75.
by uniqueuid 1y ago
I'm kind of irritated that the article doesn't say anything about probabilities. Given they have enough data, at some point it's perfectly reasonable to have cars with 5+ crashes per 12 months - just because of chance. This i
76.
by uniqueuid 1y ago
Sorry but I don't think that's right. Do you have some sources for Sociology being hit harder than, say, Psychology, in the replication crisis? I personally don't even know of a many labs replication attempt from them (Sociol
77.
by uniqueuid 1y ago
Haha sorry, I understand that reaction to quite a lot of sociological theory. Personally I find this the easiest simple mental model for reflexive modernity: "Modernity linearly accelerated change. Reflexive modernity loops back on its
78.
by uniqueuid 1y ago
Sociological theory says yes, the world is a lot more (=superlinearly) complex. https://en.wikipedia.org/wiki/Reflexive_modernization
79.
by uniqueuid 1y ago
I find experience to be a bad explanation here, because a course can absolutely teach you all these things that are discussed in the article. Sewing? The course will explain and show you, after which you gain experience by practice. Making
80.
by uniqueuid 1y ago
You have a point with democratic backsliding - but then your rights hinge on the impartiality of the judicial system (as a whole, and eventually, not necessarily individual decisions evidently). It’s pretty obvious that the legal systems ev
81.
by uniqueuid 1y ago
You’re not arrested for posting this, so that is a pretty big difference to Russia (and other authoritarian nations like China and Turkey), no? https://rsf.org/en/country/russia
82.
New desktop and foldable laptop from framework(frame.work)
12 points by uniqueuid 2y ago|0 comments
83.
by uniqueuid 2y ago
No, they are linked to beliefs (like anything else), but the canonical forms do differ a lot. Most importantly: - bayesian methods give you posterior distributions rather than point estimates and SEs - bayesian methods natively offer prior
84.
by uniqueuid 2y ago
In general you assume DAGs, i.e. non-cyclical causality. Cyclical relations must be resolved through distinct temporal steps, i.e. u_t0 causes v_t1 and v_t1 causes u_t2. When your measurement precision only captures simultaneous effects of
85.
by uniqueuid 2y ago
You are missing one crucial additional condition: - No colliders have been included in the analysis, which would introduce appearance of causality that does not exist
86.
by uniqueuid 2y ago
Yes it's great! There is also this great book on causality in ML, but it's a much heavier read: Chernozhukov, V., Hansen, C., Kallus, N., Spindler, M., & Syrgkanis, V. (2025). Causal Inference with ML and AI.
87.
by uniqueuid 2y ago
Yeah but in this case it's really a wrong way to think about it. If you have a DAG based on wrong assumptions, it doesn't matter whether you get a point estimate based on null hypothesis thinking or whether you get a posterior dis
88.
by uniqueuid 2y ago
Funnily enough that hardly matters here. Causality is a largely orthogonal problem to frequentist/bayesian - it makes everything harder, not just one of those!
89.
by uniqueuid 2y ago
"provided one dares assume ..." - that's a great quote which I'll steal in the future if you allow! Most things we learn about DAGs and causality are frustrating, but simulating a DAG (e.g. with lavaan in R) is a techn
90.
by uniqueuid 2y ago
Oh what fun to discover the horror of causality! For some areas of research, truly understanding causality is essentially impossible - if well-controlled experiments are impossible and the list of possible colliders and confounders is unkno

More ›