iiinova — where your digital self keeps evolving

Andrzej Wąsowski 🟥 @andrzej.dukla.ch.ap.brid.gy · 13/10/2026

It's enchanting how empty /bin and /usr/bin are on #nixos. (arguably /bin looks the same on arch, but /usr/bin is full there). #nerdTalk #linux

sh-5.3$ ls -l /bin /usr/bin
/bin:
total 4
lrwxrwxrwx 1 root root 73 Oct 13 14:31 sh -> /nix/store/pwxj9ycwqd117yra6czifh3xfx5p88lh-bash-interactive-5.3p9/bin/sh

/usr/bin:
total 4
lrwxrwxrwx 1 root root 66 Oct 13 14:31 env -> /nix/store/bhv6xyvqw6gjs3j488gi1z6wdv18bdpa-coreutils-9.11/bin/env

010 1

Andrzej Wąsowski 🟥 @andrzej.dukla.ch.ap.brid.gy · 11/09/2026

The probabilistic programming library #Probula in #Scala, got a new minor release today (v.0.2.3). The release includes a bunch of small improvements that slowly make probula usable: (support for vectorized variables/plates, for-comprehensions, truncated […] [Original post on social.dukla.ch]

$```scala val model = for a <- Gaussian("a")(0.0, 10.0) // prior b <- Gaussian("b")(0.0, 3.0) // prior σ <- UniformC("σ")(0.0, 3.0) // prior _ <- xs.plateD { x => Gaussian(a*x + b, σ) } // likelihood .observing(ys) // ... and data yield (a, b, σ) ```$

001 1

Andrzej Wąsowski 🟥 @andrzej.dukla.ch.ap.brid.gy · 06/09/2026

_Meet racism face-to-face._ Muzeum Karykatury in Warsaw, Kozia st., has a small temporary exhibit of Polish zines of the transformation period, roughly 1980-2010. Full spectrum, anti communist, extreme left, extreme right, and arts. There is some minimal […] [Original post on social.dukla.ch]

Meet racism face-to-face. Or: face racism.

030 3

Andrzej Wąsowski 🟥 @andrzej.dukla.ch.ap.brid.gy · 05/08/2026

A small step for the humankind, a huge celebration for my inner nerd. The most beautiful formulation of Fisher's 1922 dilution assay model in the history of humankind has landed 😊. A lot has been happening in Probula lately, but chiefly I worked on syntax to […] [Original post on social.dukla.ch]

$A piece of Scala code (Probula, an internal DSL) encoding Fisher's model, a screenshot from a neovim editor buffer. The code is as follows: ```scala import math.log val data = FisherPlated.dataK val posterior = for ρ <- UniformC(0, 2000) // prior η = (1 to K).map { i => log(ρ) - i*log(2) } // linear predictor πs = η.map { Cloglog.inverse } // (inverse) link cloglog^-1 _ <- πs.plateD { Binomial(N, _) }.observing(data) // likelihood+data yield ρ // Importance sampling inference val stats = posterior.sample(1000.sampleSize).precis() ```$

000 0