> If you vibe coding the errors are caught earlier so you can vibe code them away before it blows up at run time
You can say that again.
I was looking into the many comments for this particular comment and you did hit the nail on the head.
The irony is that it took the entire GenAI -> LLM -> vibe coding cycle to settle the argument that typed language is better for human coding and software engineering.
Sure, but in my experience the advantage is less than one would imagine. LLMs are really good at pattern matching and as long as they have the API and the relevant source code in their context they wont make many/any of the errors that humans are prone to.
>who is using iceberg with hundreds of concurrent committers, especially at the scale mentioned in the article (10k rows per second)? Using iceberg or any table format over object storage would be insane in that case
You can achieve 100M database inserts per second with D4M and Accumulo more than a decade ago back in 2014, and object storage is not necessary for that exercise.
Someone need to come up with lakehouse systems based on D4M, it's a long overdue.
D4M is also based on sound mathematics not unlike the venerable SQL [2].
[1] Achieving 100M database inserts per second using Apache Accumulo and D4M (2017 - 46 comments):
Imagine a standard based non-prorietary WhatsApp alternative that works globally and seamlessly between different messaging companies and service providers via SMS, no extra app to install and just work.
Merely 2000 words, we have a full complete book for that [1].
Joking aside, D4M has seamlessly combined spreadsheet, table, database and graph concepts based on associative array mathematics [2].
On one extreme people are going to bolt on everything on Postgresql database, and another extreme of integrating clunky disparate systems, D4M is a breath of fresh air that is based on mathematics not unlike the venerable SQL relational database concepts [3].
[1] Mathematics of Big Data
Spreadsheets, Databases, Matrices, and Graphs:
Please check this excellent LLM-RAG AI-driven course assistant at UIUC for an example of university course [1]. It provide citations and references mainly for the course notes so the students can verify the answers and further study the course materials.
[1] AI-driven chat assistant for ECE 120 course at UIUC (only 1 comment by the website creator):
I've worked on systems where we get clickable links to source documents also added to the RAG store.
It is perfectly possible to use LLMs to provide accurate context. It's just asking a SaaS product to do that purely on data it was trained on, is not how to do that.
I haven't seen it happen at all with RAG systems. I've built one too at work to search internal stuff, and it's pretty easy to make it spit out accurate references with hyperlinks
LLM foremost killer application is what I called context searching whereby it utilized RAG and other techniques to reduce hallucinations and provide relevant results in which arguably ChatGPT is one of the pioneers.
LLM second killer application is for studying for a particular course or subject in which OpenAI ChatGPT is also now providing the service. Probably not the pioneer but most probably one of the significant providers upon this announcement. If in the near future GenAI study assistant can adopt and adapt 3 Blue One Brown approaches for more visualization, animation and interactive learning it will be more intuitive and engaging.
Please check this excellent LLM-RAG AI-driven course assistant at UIUC for an example of university course [1]. It provide citations and references mainly for the course notes so the students can verify the answers and further study the course materials.
[1] AI-driven chat assistant for ECE 120 course at UIUC (only 1 comment by the website creator):
It will be very interesting to see the data for the same car that has many powertrain versions for example the Lexus UX with the UX 200 (ICE), UX 300h (hybrid) and UX 300e (EV) to test which one the best and the worst in term of brake dust residue.
My hypotheses is that for brake dust residue the best is hybrid, 2nd will be ICE and the 3rd will be EV. This is due to the fact that the EV version has at least several hundreds kg extra weight (about 400 kg extra), that makes the brake dust residue comparable to ICE if not worst based on the approximately 30% extra vehicle weight for the battery. The hybrid however only has approximately 5% more weight or extra 80 kg different compared to the ICE version.
I think buyer demographics are gonna play hugely into it. Some makes and models are highly popular among the drivers who are on the low side of the bell curve and basically never hit the brake when not stopping because they're almost never coming upon slower traffic. Some makes and models are highly popular on the other side of the peak of the bell curve where the drivers are always hitting the brake way more than the median or average. An ICE Tacoma may very well use way less brake than a EV Altima because the venn-diagram of people who drive like a bat out of hell and the people who buy Tacomas is approximately two circles.
> that makes the brake dust residue comparable to ICE if not worst based on the approximately 30% extra vehicle weight for the battery.
Did you miss pretty much all data on EV brakes, notably that they get used so little they’ll rust to slick and manufacturers have to implement de-rusting cycles to ensure they can actually do something? Your hypothesis is nonsensical on its face. Calling it a hypothesis is insulting. Even to flat earthers.
Fun facts the venerable Border Gateway Protocol or BGP is using distant-vector algorithm for global Internet routing calculation based on Bellman-Ford dynamic programming approach due to its scability in updating the routing table [1].
For intranet routing however shortest path algorithm based on Dijkstra approach underlies the OSPF and IS-IS protocols.
In the old days distant-vector protocol like Routing Information Protocol protocol or RIP is also used but it's now becoming obsolete and humorously known as Rest In Peace protocol.
Distant-vector now is making a comeback for intranet in the form of mobile ad hoc and mesh networks (guerilla networks for routing without infrastructure) with Ad hoc On-Demand Distance Vector Routing protovol or AODV [2]. This setup can be very beneficial for disaster and also for mobile group or platoon communication.
The difference is that if the NSA has physical access to my phone, I'm probably aware of it. It makes routine fishing expeditions across broad populations much harder and more expensive, as well as easier to oppose.
If they can fish remotely and automatically, accountability goes completely out the window.
You can say that again.
I was looking into the many comments for this particular comment and you did hit the nail on the head.
The irony is that it took the entire GenAI -> LLM -> vibe coding cycle to settle the argument that typed language is better for human coding and software engineering.