HN user
Business, Tech, Supply Chain, ML, AI.
How does one know whether they actually learnt physics this way. Seems a lot of reading and structure, that's great. Not convinced about whether there is a change in understanding. Typical university/college classes have assignments, quizzes, exams etc.. don't see the equivalent.
I laughed at this "but the code complexity became too much for me, and I abandoned the project." even Terry Tao finds some code too complex to write.
Really bullish on LLMs expanding code development by a very large group of people who are really smart in some domain but could not get into 'coding'.
Rewrite in C is the new Rewrite in Rust.
So, and iPad with an attached keyboard.
If it is this easy, surely the trend is Rust output being an intermediate pass of the LLM super compiler. A security subset if you will (like other kinds of optimization), it will move from Rust specs to some deeper level of analysis and output the final executable. Some brave souls will read the intermediate Rust output (just like people used to read the assembler output from compilers) but the LLM super compiler will just translate a detailed English like spec into final executables.
Good. How does it compare to cuOPT which seems to be 10-5000x faster than existing solvers for Vehicle routing if you believe the marketing.
Not Rust because you won't find anyone to maintain code. Rust is only for rewriting existing code in other language not maintaining it. C++ perhaps because of number of people who already know it, but only for performance sensitive areas. Also, lot of training data. Other than that, usual suspects, Java/C# for enterprise and node/Python for web/DS.
LOL, the blog gives a lot of detailed reasons, even summarizes it [1] and but some random stranger gives an outdated opinion from the '90s, which is not even wrong just plain humorous. If slave labor, how come everything else is also not so cheap.
[1] Virtually all the major mechanisms that can drive efficiency improvements — improving technology and overlapping S-curves, economies of scale (including geometric scaling effects), eliminating process steps, reducing variability and improving yield, advancing towards continuous process manufacturing — are on display hereAndy is probably the only person who adores Larry Ellison (Oracle) unironically.
Seems like a bug, it's pulling in all the jobs that have either vision or pro in their text (which is probably everything). Need to put "vision pro" to filter and even select the product (Apple Vision Pro) to get about 74 jobs.
This is a good resource, however for about 99.99% of people, you are most likely to just use a foundation model like ChatGPT, Claude, Gemini etc. so this knowledge/training will get you neither here or there. I would suggest you look into another Karpathy's video -- Deep Dive into LLMs like ChatGPT.
LOL, this is the list to keep in your head for this so called "manual". Best of luck of those who will work through this. BTW, Karpathy made that comment in 2025 not 2024.
Morphability - natural language as morphable code
Abstraction - tasks become reusable commands
Recursion - stack abstractions for leverage
Internal Consistency - prevent system drift
Reproducibility - crash-resilient design
Morphic Complexity - recognize over-engineering
E2E Autonomy - measure actual capabilities
Token Efficiency - maximize work per token
Mutation & Exploration - controlled self-improvementSeems you are paying the Dell tax of 15%. The same setup is $4K from NVidia, Lenovo and $3K for 1TB at Asus.
https://www.dell.com/en-us/shop/desktop-computers/dell-pro-m...
You just need to master one package managed in depth and you will get what you really want with Modern C++.
Just like San Francisco and Dallas/Texas (from his article) are very different in the US, we should expect lot of differences in Europe (as others mentioned, he clubs UK with EU). Housing is a general problem for all major cities though, not sure why you think it is unique to London in the whole continent. Stockholm, Paris, Dublin, Lisbon to name a few, are pretty bad for housing in their own unique ways. Certainly shouldn't be "breaking your brain".
Great summary of the year in LLMs. Is there a predictions (for 2026) blogpost as well?
Seems very detailed and comprehensive. Did I miss it, but was there a performance comparison to the PyTorch version at the top?
Three things:
1. The Rise: 2005 - 2010 Google hired Guido van Rossum in 2005 (stayed on for seven years) and gave corporate blessing that made everyone comfortable with moving from Perl to Python. It was seen as the language of scientists and smart people so a lot of people working in misc. languages like Fortran, MATLAB, Perl moved here. To remove the speed issue the official Google mantra was "Python where we can, C++ where we must". AI heavy weights like Peter Norvig (I think he was the chief AI scientist at one point and co-author of the famous AIMA book), promoted Python to be an acceptable Lisp.
2. Near Death: 2010-2015 Python almost died due to self inflicted wound from the 2 -> 3 transition and there was a good chance it would have gone nowhere like many languages before. Guido also moved away from Google and Google seemed to have shifted it's attention to Golang (apart from the standard C++ and Java). BTW, Python's dominance was not seen positively within Google hence they stopped actively promoting it. For ex. a leaked transcript from Eric Schimdt had him saying this
So another definition would be language to Python, a programming language I never wanted to see survive and everything in AI is being done in Python.
3. Resurrection: 2015-Now Data science and ML took off and Python was right there thanks to the initial sponsorship from Google and ecosystem of scientists and engineers who were familiar (including working in the two-language mode). There was no language that could rival at this point.
Most of the syntax, power considerations etc.. are side shows as most scripting languages just tap into very powerful libraries written in c/c++/fortran or wrappers around shell. Doubt that distinguishes Python to the point where it has become so dominant.
Confluent was trading at less than 50% of its IPO price when IBM made the offer. The stock and the company has been going sideways for several years now, keeps growing revenues but loses even more as most of it is in Sales and Marketing. In which world is this seen as some sort of extraordinary company that will get sabotaged by IBM. Seems Confluent management knows the writing on the wall, IBM will clean up (fire a bunch of sales and management guys) and make this a workable business. It will seem brutal for some Confluent guys but that's because their business is broken; and only someone from outside can come in and fix it as the current senior management cannot.
IBM has been around for over a hundred years, maybe they know a thing or two about running a software business :-)
Kind of strange take as though unique to software. Every sector that is large has issues since ambitious projects stretch what can be done by the current management and organizational practices. All software articles like these hark back to some mythical world smaller in scope/ambition/requirements. Humanity moves forward
* Construction and Engineering -- Massive cost overruns and schedule delays on large infrastructure projects (e.g., public transit systems, bridges)
* Military and Government -- Defense acquisition programs notorious for massive cost increases and years-long delays, where complex requirements and bureaucratic processes create an environment ripe for failure.
* Healthcare -- Hospital system implementations or large research projects that exceed budgets and fail to deliver intended efficiencies, often due to resistance to change and poor executive oversight.
Python is nothing without it’s batteries.
I think this is the key, it can do impressive stuff but it won't be fast. For that, you have to put in a NVidia data center / AI Factory.
He is very enthusiastic about new things but even he struggled (for ex. the first link is about his experience OOB with Sparq and it wasn't a smashing success).
Should you get one? #
It’s a bit too early for me to provide a confident recommendation concerning this machine. As indicated above, I’ve had a tough time figuring out how best to put it to use, largely through my own inexperience with CUDA, ARM64 and Ubuntu GPU machines in general.
The ecosystem improvements in just the past 24 hours have been very reassuring though. I expect it will be clear within a few weeks how well supported this machine is going to be.Seems this is basically DGX Spark with 1TB of disk so about $1000 bucks cheaper. DGX Spark has not been received well (at least online, Carmack saying it runs at half the spec, low memory bandwidth etc.) so perhaps this is way to reduce buyers regret, you are out only $3000 and not $4000 (with DGX Spark).
Rust is a worse C++ with modern tooling.
How will Fivetran and dbt who are detested for being overpriced and underfeatured in the segment they are supposed to be good at (ETL/ELT) be taking on being a datalake? That's orders of magnitude more complex in engineering and operating and they have no experience. This is really a play to consolidate, get rid of duplicate functions and provide a better experience to customers.
Well said, the point of doubly linked lists is not that it is the be all and end all of all data structures. You can possibly workaround that. The real issue is a real program has much more complex data structures, how can you possibly make it work or refactor easily when a simple thing like DLL's are causing so much grief. So, yeah Rust really is unproductive in a massive way. Just useful in small parts where you might really need memory safety (that too, not sure why you wouldn't chose another language).
Where are you seeing proprietary compilers freezing support after C++17. C++20 support is almost green across the board (even Cray) and C++23 is getting there. As usual there is a lag, for ex. when nvcc will declare C++23 support and whole swath will turn green. Right now it only supports C++20 (but you can mix in C++23 files and link them together) etc.. I see most C++ projects make C++20 a minimum. (not counting someone who wants to keep running C++98).
Contributors
claude Claude
Interesting given Microsoft’s history with OpenAI
Did nobody ask why you would want to reuse contact lens fallen on some grimy ground in the night? Keys, coins, bills, even glasses ok, but putting back the contact lens into your eyes seems dangerous. I thought some kid might have spoken up :-).