Varun (@typesanitizer) header
Varun (@typesanitizer) profile picture

I am once again asking you to expect more from the tools you use every day. OK: English, हिंदी, ગુજરાતી; limited 日本語, 中文. Avatar by @evatrice. He/Him.

Amsterdam, The Netherlands
Joined May 2013
Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)Photo by Varun (@typesanitizer)
Give a man a counterexample, and he will question his instincts for a day. Teach a man to find counterexamples, and he will be trapped in mental agony forever wondering if his inability to find a counterexample is a counterexample for his skillfulness in finding counterexamples.
1
13
76
Thanks for having me on the podcast! I had fun talking about various topics like software performance, the changing expectations around dev tools, and the enduring value of understanding the details.
Race conditions used to scare engineers into getting concurrent code right the first time. Varun Gandhi worries that LLMs change that math. If a model can debug a race condition later, teams may get comfortable shipping more bugs now.
52
A naive understanding of formal verification might be something like, "Well, if I do it in Lean, then it can't have bugs." WDYM "do it"? WDYM "can't have bugs"? This is entirely the question.
2
6
53
4,321
It's game night. You're playing Set, blithe and buoyant within an edenic oasis where functions are total and least and greatest fixed points are distinct. In the tarpits below I play DCPO⊥, grizzled by the uncaring realities of a nonterminating world. We are not the same.
1
5
43
The more I dig into the fundamentals of programming education, the more I realise we've had a truly profound failure to visualise both the structure of programs themselves as well as the data of running programs. Likely we've selected for people who can mentally visualise.
39
35
10
939
33,173
Reminder that it takes just as much effort (time, money, staff, etc.) to build the wrong thing as it does the right thing. Problem discovery is under appreciated as a practice in software engineering, especially when compared to problem solving.
22
702
45
1,523
Trying this for now. Let's see. 🤞
+1 Don’t fear muting and unfollowing people who “changed.” Twitter only works well when you curate your experience. For example, I’ve muted “AI” and a few other related keywords, which helps bring the noise down
4
136
Been working on a blog post which cites 'Nobody Ever Gets Credit for Fixing Problems that Never Happened' by Repenning and Sterman; it's a bit of a struggle to pick out specific bits, because the paper is full of these nuggets of wisdom.
6
142
I guess this is the next level after "The monetization strategy is that my bank account has 3 commas mate"
Jensen Huang on Ezra this week:
1
125
Me: I know a little bit about ordinal numbers, can you explain forcing to me? ... (few messages later, where I ask for a simpler formulation of a countable model of ZFC)
1
3
130
Inspired to try this based on xitter.gq/DRMacIver/status/21017…
An irony of "open exposition problem" being in the spotlight due to AI generated mathematics is that over the last couple of days I've been getting ChatGPT to tutor me in Forcing, the topic which coined the term and which was the first place I hit a wall in mathematics.
46
Proposal: GitHub contribution color changes to ochre the more slop you contribute
5
117
The layers in this meme. Remarkable.
This is my new shortest “what is Jev” explanation
105
Do say hi if you're here at JJ Con / Git Merge in Lisbon this week 👋🏽
4
103
Varun retweeted
New blog post is out! A Zoomable Equal-Area Map. Equal-area maps are now recommended by the UN, but devs usually prefer the Mercator projection. This new map combines both map types with a very simple mathematical formula. benjoffe.com/map
1
1
4
328
Proposal: if you submit a drive-by perf/feature PR to an unfamiliar part of the codebase, that should be preceded or succeeded by at least 2 PRs which either simplify the code, fix bugs, or improve test coverage
1
4
187
I've been crystalizing some advice for first-year college students: "Advice to the New College Student in the AI Era". Thesis: "the purpose of going to college is to emerge as close to middle-aged as possible". docs.google.com/document/d/e…
14
41
12
257
17,133
I don't usually participate in these but trying this for a change.
if you see this share some of your favourite painting
1
3
329
"What is hard about software engineering continues to be hard. Actually saying no is materially harder if the machine keeps making pull requests. [..] saying no is still really an important part about the whole thing." -- From @mitsuhiko in youtu.be/9UAxrdcDjjU?si=3Ult…
1
129