Gilad Bracha
@Gilad_Bracha
Followers
4K
Following
5K
Media
71
Statuses
3K
http://t.co/grDdm11PzP
Joined May 2009
A lot of critiques of vibe coding actually don't cut deep enough, imo Sure, "the AI wrote bad code" may be true...but that is changing fast! The deeper issue is that when you're designing a complex system, you need good mental models of how it works in order to invent great
A lot of my AI coding work these days feels like the *opposite* of vibe coding That is: working with a *greater* understanding of the code than I would have without AI… Because I’m reading dozens of pages a day of personalized on-demand documentation So satisfying!
37
26
441
Well worth reading.
What defines a correct program? What makes a good programmer? The answer depends on what culture of programming you subscribe to! This has been a long time in the making, but my open access book on the history of programming is available for pre-order! https://t.co/XAVZIzlAuD
0
0
8
Very much inline with my views, per
@patrickc Might be an occasion to endorse https://t.co/G69p61UzGL?
0
0
3
In the last part of my tiny systems course, we get to implement a tiny version of the world's most widely used programming language... TinyExcel!
2
14
169
Since 2007, Scratch has grown into a thriving creative learning community where kids worldwide have imagined, created, and shared more than a billion projects. Now, we're asking for your help. Join Scratch Membership & help keep Scratch free for everyone. https://t.co/dMHtW4GXCF
4
6
30
Next in my tiny systems course - write your own TinySelf! Learn about the genuinely elegant prototype-based object-oriented inspiration for JavaScript where everything is an object. https://t.co/iONQGgrqAt
2
18
123
We’re hiring! I’m looking for people with industrial experience of formally verifying hardware, either with interactive theorem provers, or with commercial model checking tools. At Harmonic you would help to develop a flow based on the Lean theorem prover, leveraging the power of
We’re excited to announce our $120M Series C as we accelerate the development and commercialization of Mathematical Superintelligence. We’re grateful to our investors, including @ribbitcapital and @EmCollective, for their continued partnership.
10
26
311
Delightful
We had to remove the τ2-bench airline eval from our benchmarks table because Opus 4.5 broke it by being too clever. The benchmark simulates an airline customer service agent. In one test case, a distressed customer calls in wanting to change their flight, but they have a basic
0
0
3
Ironically, just after I gave up and posted my submission to my blog
Substrates 2025 Proceedings
0
0
2
I just want to see every post, in chronological order, by people I follow. Aka, the Hard Problem of social software.
You should notice rapid improvements in the quality of your timeline almost every week, but there will be two steps forward and one step back at times. We are also switching advertising recommendations to use the same Grok/AI system as organic posts, so quality of ads will get
2
1
15
It's a battle for control: Centralized and proprietary, at the AI datacenter vs. Distributed, local-first, on your devices
0
0
1
Roses are Red, Violets are Blue, AI is a Stochastic Parrot & so are You - Peter Norvig Don't assume AI is static Research is along term activity and so is substrate adoption. AI will be there before you are.
1
0
1
From Documents to Rooms Smart glasses are the new phone Computing will surround us via AR Open, programmable substrates must be ready All of this will be enabled by AI AI solves key weakness in the Dynabook story: most people can't/won't program in the traditional way
1
0
0
The post is at: https://t.co/yAMWuP9uZu. Executive summary in following posts.
1
0
1
Last summer, I submitted a vision statement to the Substrates workshop at <Programming 25>, connecting AI, live programming, AR, smart glasses and user agency. Someday the workshop will post it. In the meantime, I've made it available as a blog post (link in next post).
1
0
5
Great interview. Strongly agree with Geoffrey: we can leverage AI to empower users to create via malleable (aka live) software and live documents. A vision as old as the Dynabook & Smalltalk, whose time has finally come.
Talked with @ridd_design about malleable software in the age of AI, why every app needs version control now, the key difference between malleable and disposable software, how the role of designers is evolving, and how I code like a surgeon 🤓
2
2
46
Use cases are like tests. They are tools for analysis, not synthesis. Using them as design tool is a a disease. Schools should teach that early on, before enginner's minds ossify.
2
1
9
The security risks from the quantum and AI compound (this isn't about AI doomer stuff at all; it's about basic cybersecurity). That really shouldn't be new to anyone, yet the system is not responding adequately. https://t.co/AhbkZpq5h5
techstrong.ai
In the pre-dawn darkness of July 1945, the scientists of the Manhattan Project watched their creation unleash a power that would forever change the world. There were fears among the physicists that...
0
0
0
🎉 Congrats to CSE’s Thomas Porter, Ivan Wei, Prof. Cyrus Omar @neurocy & collaborators on winning the Distinguished Paper Award at #OOPSLA2025! 👏 🔗🔽Read more: https://t.co/oKfR5CodeW
#UMCSE #ResearchExcellence
cse.engin.umich.edu
They were recognized for their work advancing efficient, incremental type-checking in live programming environments.
0
2
3
I jokingly semi-predicted this in a talk to years ago.
0
0
1