iiinova — where your digital self keeps evolving

Alecs P. Hacker @bisimulation.bsky.social · 03/10/2026

Mom: Oakland is considered one of the most dangerous cities in the usa

150

Alecs P. Hacker @bisimulation.bsky.social · 06/09/2026

Brother i was having a good fucking day why have you decided of ruining it

060

Alecs P. Hacker @bisimulation.bsky.social · 31/08/2026

On my phone it shows up as lake ontario

000

Alecs P. Hacker @bisimulation.bsky.social · 30/08/2026

Now i see why university is so expensive here

060

Alecs P. Hacker @bisimulation.bsky.social · 23/08/2026

"math/computer science researcher and internet personality" im neither

AI Overview Here is a tweet written in the distinct voice and thematic style of the math/computer science researcher and internet personality KleeneAlgebra:"the absolute hubris of writing code without a formal proof of correctness. just vibes and unit tests. you are praying to a god that is not listening to you. look at the state transition matrix. it is weeping."🧱 Style AnalysisAcademic Despair: Melding advanced theoretical computer science (formal verification, matrices) with existential dread.Vibes vs. Math: Contrasting modern tech-bro slang ("just vibes") against rigid mathematical structures.Poetic Nihilism: Treating software bugs or engineering oversights as cosmic, tragic failures.Would you like me to draft another tweet focusing on a specific computer science topic (like type theory, category theory, or Rust vs. C) or adjust the level of absurdity?

010

Alecs P. Hacker @bisimulation.bsky.social · 23/08/2026

I'm starting to think that schooling is literally illegal in this country

Alessio: I am politically liberal which is why I'm sending this message. Our country is kidnapping people and putting them in detention centers all over the country if they are not american. It doesn't matter the reason. They're pulling people out of colleges, out of stores, etc. If you don't have a social security card, do you have a Visa? I've got someone else interested in my unit starting August 29th. I want to be fair but I don't want to be part of someone getting kidnapped in our country. That would make me physically ill.

010

Alecs P. Hacker @bisimulation.bsky.social · 17/07/2026

Can we stop adding low quality LLMs to everything?

Original:
Ora hai accesso promozionale alla funzione AI in Fogli

Genera, riepiloga o suddividi in categorie un testo per campagne email, annunci, revisioni e altro ancora quando digiti =AI

Troll language:
You now have promotional access to the AI feature in Sheets.

Generate, summarize, or categorize text for email campaigns, ads, reviews, and more when you type =AI.

250

Alecs P. Hacker @bisimulation.bsky.social · 22/06/2026

We do not need dependent types anymore

> USER: attached: paper.pdf message: Find holes in the proofs
> CLAUDE: *Thinking*

1160

Alecs P. Hacker @bisimulation.bsky.social · 17/06/2026

No mathpartir no problem

$\frac{\vdash \cancel{A} \land B}{\vdash \cancel{A}}$

190

Alecs P. Hacker @bisimulation.bsky.social · 11/06/2026

AGI does not know about strict proof irrelevance

Me:
Translate this in rocq

inductive N : Prop
| Z : N
| S : N → N
open N

theorem chicken : Z = (S Z) := by simp

GPT:
The direct Rocq/Coq translation is:

Inductive N : Prop :=
| Z : N
| S : N -> N.

Theorem chicken : Z = S Z.

However, unlike the Lean snippet, this theorem is not provable in Rocq. In fact, Z and S Z are distinct constructors of the inductive type N, so Rocq rejects any attempt to prove it.

You can prove its negation:

Inductive N : Prop :=
| Z : N
| S : N -> N.

Theorem no_chicken : Z <> S Z.
Proof.
discriminate.
Qed.

The Lean code

theorem chicken : Z = (S Z) := by simp

would not be accepted by Lean either (assuming standard definitions), because simp cannot prove an equality between different constructors. It would fail with a goal equivalent to:

⊢ Z = S Z

which is false.

170

Alecs P. Hacker @bisimulation.bsky.social · 25/05/2026

It's very hard to take a picture alone

The lambda calculus omega term tatted on an arm

4111

Alecs P. Hacker @bisimulation.bsky.social · 15/05/2026

Coff

$synthesize (TmApp tm1 tm2) = do (ty1, tm1) <- synthesize tm1 (dom, bcod) <- ty1 & fix \rec -> \case -- (Alecs) NOTE: We can't directly match on TyArr as it could have -- been refined TyArr dom bcod _ -> pure (dom, bcod) TyRef base _ _ -> rec base _ -> throwLocalError $ FunctionExpected tm1 ty1$

130

Alecs P. Hacker @bisimulation.bsky.social · 13/05/2026

One of the advantages of living in a big city

0180

Alecs P. Hacker @bisimulation.bsky.social · 20/04/2026

Clearly worth the 300 lines of configuration

Screenshot of a latex listing using the cattpuccin latte theme

190

Alecs P. Hacker @bisimulation.bsky.social · 10/04/2026

Should Liquid Haskell be AI powered?

Mail from slack about some AI crap that has the line "An Admin/Owner must first enable AI for Liquid Haskell"

1140

Alecs P. Hacker @bisimulation.bsky.social · 03/04/2026

My town got so gentrified and colonized by the Germans and Dutch that now we have "dutch specialities" and then it's rat sausages

Sign that says "specialità olandesi frikandellen" - "dutch specialities frikandellen"

030

Alecs P. Hacker @bisimulation.bsky.social · 16/02/2026

Why are looks maxxers obsessed with separation logic?

000

Alecs P. Hacker @bisimulation.bsky.social · 27/01/2026

The famous PhD level intelligence is unable to read a label in LaTeX

$Query: `Complete \label{pf:tm-per-semantics} in the same way as \label{pf:tm-per-bot-semantics}` diff where the ai modified a theorem with label pf:ty-per-semantics$

170

Alecs P. Hacker @bisimulation.bsky.social · 16/01/2026

Pretty sure this guy knows about it

Book: Theories of programming languages by Reynolds

110

Alecs P. Hacker @bisimulation.bsky.social · 13/01/2026

The emacs experience is rewriting the whole package from scratch with advices

$(use-package z3-mode :init :config (defvar alecs/z3-solver-cmd (concat (executable-find "z3") " -in")) (defvar alecs/cvc5-solver-cmd (executable-find "cvc5")) ;; By default this package loads z3 (defvar alecs/smt-solver-region-command alecs/z3-solver-cmd) (defun alecs/z3-mode-use-cvc5 () "Switch Z3 mode to use CVC5 as the solver." (interactive) (setq alecs/smt-solver-region-command alecs/cvc5-solver-cmd) (setq z3-solver-cmd (executable-find "cvc5"))) (defun alecs/z3-mode-use-z3 () "Switch Z3 mode to use Z3 as the solver." (interactive) (setq alecs/smt-solver-region-command (concat (executable-find "z3") " -in")) (setq z3-solver-cmd (executable-find "z3"))) (defun alecs/z3-execute-region-advice () (shell-command-on-region (if (region-active-p) (region-beginning) (point-min)) (if (region-active-p) (region-end) (point-max)) alecs/smt-solver-region-command)) (advice-add 'z3-execute-region :override #'alecs/z3-execute-region-advice) :mode ("\\.smt2\\'" . z3-mode))$

020

Alecs P. Hacker @bisimulation.bsky.social · 10/01/2026

I ran it more but alle it did was lying on the progress and ignore what we did in the paper

260

Alecs P. Hacker @bisimulation.bsky.social · 06/01/2026

Now I can start doing actually useful stuff like working on implementing the changelist

$data Ty : Unit = iota : Ty unit | arr : Ty unit -> Ty unit -> Ty unit in strat Val : Ty unit = Viota : Val Ty.iota | Varr : (t1 : Ty unit) -> (t2 : Ty unit) -> (Val t1 -> Val t2) -> Val (Ty.arr t1 t2) in data Ctx : Unit = empty : Ctx unit | cons : Ctx unit -> Ty unit -> Ctx unit in data CtxTy : Unit = mk : Ctx unit -> Ty unit -> CtxTy unit in data Ref : CtxTy unit = here : (ctx : Ctx unit) -> (ty : Ty unit) -> Ref (CtxTy.mk (Ctx.cons ctx ty) ty) | there : (ctx : Ctx unit) -> (ty1 : Ty unit) -> (ty2 : Ty unit) -> Ref (CtxTy.mk ctx ty2) -> Ref (CtxTy.mk (Ctx.cons ctx ty1) ty2) in data Lam : CtxTy unit = lam : (ctx : Ctx unit) -> (ty1 : Ty unit) -> (ty2 : Ty unit) -> Lam (CtxTy.mk (Ctx.cons ctx ty1) ty2) -> Lam (CtxTy.mk ctx (Ty.arr ty1 ty2)) | app : (ctx : Ctx unit) -> (ty1 : Ty unit) -> (ty2 : Ty unit) -> Lam (CtxTy.mk ctx (Ty.arr ty1 ty2)) -> Lam (CtxTy.mk ctx ty1) -> Lam (CtxTy.mk ctx ty2) | var : (ctx : Ctx unit) -> (ty : Ty unit) -> Ref (CtxTy.mk ctx ty) -> Lam (CtxTy.mk ctx ty) in data Env : Ctx unit = emptyEnv : Env Ctx.empty | extendEnv : (ctx : Ctx unit) -> (ty : Ty unit) -> Val ty -> Env ctx -> Env (Ctx.cons ctx ty) in let lookupRef = rec r : (ctx : Ctx unit) -> (ty : Ty unit) -> Ref (CtxTy.mk ctx ty) -> Env ctx -> Val ty . \ctx . \ty . \ref . \env . case ref as _ in Val ty of | Ref.here k z -> case env as _ in Val ty of | Env.extendEnv _ _ v _ -> v end | Ref.there ctx _ _ ref -> case env as _ in Val ty of | Env.extendEnv _ _ _ env -> r ctx ty ref env end end in let eval = rec r : (ctx : Ctx unit) -> (ty : Ty unit) -> Lam (CtxTy.mk ctx ty) -> Env ctx -> Val ty . \ctx . \ty . \term . \env . case term as _ in Val ty of | Lam.lam _ ty1 ty2 body -> Val.Varr ty1 ty2 (\v1 . r (Ctx.cons ctx ty1) ty2 body (Env.extendEnv ctx ty1 v1 env)) | Lam.app _ ty1 ty2 fn arg -> case r ctx (…$