- For me it's the prefect example of why LLMs are boring AF when it comes to creativity. Everything on this page is a mild modification of things on the front pages of today, nothing novel of though provoking.
Hey AI please create art, and it gives you a hue shifted Mona Lisa. I find that supremely boring.
- Our viewpoints don't seem that far apart and thanks for the nuanced take. Personally I believe we know that technology can't fix this by definition because the problem is of social, cultural and economic nature. Our lifestyles are woefully incompatible with a 100k year horizon, even a 100 year horizon in many areas. Our perception of wealth depends on never ending growth, our welfare systems depend on never ending growth, our economies depend on never ending growth. It seems implausible to the point of impossibility that our economies can grow forever [1]. Technology is good at reaching goals e.g. going to the moon is unlikely without science and technology. But in this case the problem is the goal itself. Technology won't motivate us to let go of our conveniences.
[1] https://tmurphy.physics.ucsd.edu/papers/limits-econ-final.pd...
- It would be great if we could engineer our way out of this situation, but we can't. For many years I strongly believed in our cleverness, after all I was clever and in the narrow domain I worked in - tech - cleverness was enough to overcome most issues. So why not human climate change?
In Tom Murphy's words:
> Energy transition aspirations are similar. The goal is powering modernity, not addressing the sixth mass extinction. Sure, it could mitigate the CO2 threat (to modernity), but why does the fox care when its decline ultimately traces primarily to things like deforestation, habitat fragmentation, agricultural runoff, pollution, pesticides, mining, manufacturing, or in short: modernity. Pursuit of a giant energy infrastructure replacement requires tremendous material extraction—directly driving many of these ills—only to then provide the energetic means to keep doing all these same things that abundant evidence warns is a prescription for termination of the community of life.
- This reads like a badly done, sponsored hype video on YouTube.
So if we look at what NVIDIA has to say about NVFP4 it sure sounds impressive [1]. But look closely that initial graph never compares fp8 and fp4 on the same hardware. They jump from H100 to B200 while implying a 5x jump of going with fp4 which it isn't. Accompanied with scary words like if you use MXFP4 "Risk of noticeable accuracy drop compared to FP8" .
Contrast that with what AMD has to say on the open MXFP4 approach which is quite similar to NVFP4 [2]. Ohh the horrors of getting 79.6 instead of 79.9 on GPQA Diamond when using MXFP4 instead of FP8.
[1] https://developer.nvidia.com/blog/introducing-nvfp4-for-effi...
[2] https://rocm.blogs.amd.com/software-tools-optimization/mxfp4...
- I like to call them slow-AI. They are paperclip optimizing AIs. No single component wants the larger outcomes, yet they happen. These slow-AIs are terraforming our planet into a less habitable one in order to make GDP number go up, at any cost.
- Ah yes, the if we change something that would mean we would have to change everything and that would be absurd, so we can't change anything ever argument. Classic.
- Somewhat disheartening to see all the arguments here saying it doesn't matter how we name stuff. Lived language creates reality.
There are groups of black people that complained about the terminology, repeatedly [1]. So isn't it arrogant as a white dude to say, this thing hurts you, and we could easily change it, but fuck you we won't do it because we have always done it that other way?
Some easy no effort replacements:
- master-slave -> leader-follower
- blacklist -> denylist
- whitelist -> acceptlist
[1] https://en.wikipedia.org/wiki/Master%E2%80%93slave_(technolo...
- What a waste of time article. There is plenty of debunking of the points already so I won't repeat them.
Yes C++ has many many issues, and yet most of the biggest most complex software systems in existence are written in it. I don't see the browsers, OSs, DAWs, AAA games etc. being implemented in Go. Rust is a real improvement for these areas, with a proven track record of making software more reliable and maintainable.
If the author intended to say "Rust is not the best programming language for every problem" they should have said that but it would have been worth an article as much as "Screwdrivers are not the best hammers".
- Seems like trying to solve poor performance with caching, not that all caching is bad, but it's often more rewarding to make the thing so fast it doesn't need caching if you have control over the code in question, which they very much do in this case.
- Because I've met several homeschooled adults over the years, and talking to them that's something most of them had in common when explaining the impact it had on their life. Looking for more objective data I found this one source that seems to be written by people not already convinced of the desirability of homescooling [1], forgive me for being skeptical of the objectivity of places called "national home education research institute". Overall it paints a more positive picture than I had expected, but also highlights it's limitations.
[1] http://hks.harvard.edu/sites/default/files/Taubman/PEPG/conf...
- I don't think it'll need draconian enforcement, parental controls on iOS and Android can co-exist with Linux. Having the option to enable specific filters on a client side and requiring a pass-code or OS level permissions to change them seems like a realistic way to tackle this that doesn't end in dangerous government power concentration.
As always with security, perfect is the enemy of good. A good set of hard to change - for children - client-side filters would do wonders in terms of real improvement. As much as I'm tired of the LLM hype, they might actually be a good fit for such tasks.
- Seeming mature to an adult isn't the thing in question though, is it? Not feeling or appearing awkward when interacting on their own in their 20s is what is being criticized. The anecdotal evidence you present doesn't include home schooled children in their 20s as far as I can tell.
- > And more importantly, we can provide a better environment for them to mature socially.
Citation needed.
Every perspective I've heard personally - and mirrored in comments here as well - from the non parent side of things, is quite negative in terms of learning how to behave and socialize with your peers. To you the children might seem polite and servile, and you might see this as something positive - as you state in another comment - but you are likely setting them up for life of social awkwardness and ostracization.
- Quoting their own video:
> You can verify your code quality at a glance, then ship with absolute confidence.
Proclaiming absolute confidence after a glance leaves me with scant confident in the merit of the confidence.
- > Measles, a dangerous illness that for decades has rarely infected Canadians, is back – and spreading. [...] Robert F. Kennedy, Jr., left, now the United States Secretary of Health and Human Services, stands with protesters in Olympia, Wash., in 2019, opposing a bill to tighten measles, mumps and rubella vaccine requirements for school-aged children.
Reading this, it's a challenge to feel empathy. Everyone deserves some degree of empathy, idiots too. Yet this topic seems so needlessly self inflicted. Maybe it's a more nuanced topic than I'm aware of, is there a strong argument against vaccination?
- A true human supremacist society - I know controversial term but it fits way too well. You and I live in it and it sucks how many people think like that.
Life is a co-op, winning does not mean being the last ones left standing.
- One of the best engineers I know works for [1] Woven by Toyota, Inc. They do care about getting software right, arguably more so than the german automotive sector. Don't get me wrong they all want good software, but wanting good software and putting the right systems in place while resisting the urge to chase the next hype wave are quite two different things.
- You can't do remote attestation without something like a TPM.
Let's compare these scenarios:
A) TPMs are optional and 30% of users have them. A bank is thinking about requiring remote attestation to use their services. Since they'd lock out 70% of users they decide to not do it.
B) TPMs are mandatory and 90% of users have them. A bank is thinking about requiring remote attestation to use their services. Since they'd only lock out 10% of users they decide to do it.
And banking is the nice example here. Refusing to serve a site if the user is using an ablocker is very much in the interest of powerful players in the space, see WEI. Every platform that has wide spread TPM adoption, namely Android and iOS have shown that they will abuse them for anti-consumer purposes sooner or later. We are talking about Microsoft here, the current and past poster child for anti-consumer decisions.
I hope that explains why making TPMs blanket available introduces new risks to sovereign computing.
- The way I understand it yes. Latency will be dictated by going to the leader node and replication to enough followers. The CAP theorem dictates that if you want strong consistency the way they offer you have to talk to a quorum of nodes IIRC N/2 + 1. Paxos and Raft also work that way.
Apparently there is a game about it https://tigerbeetle.com/blog/2023-07-11-we-put-a-distributed...
- > Kent lost my support in the final weeks by constantly shouting that he was a champion for his users while doing the exact opposite of what I wanted him to do.
Honestly I don't think it's just you, I'm the same boat. I find it hard to believe that the entirely predictable outcome of the current situation is what more than 1% of users wanted.
- Well put, the metric is cherry picked to further the narrative.
- Regarding initialization, if one wants portable code that works for more than one machine+compiler version, it's advisable to program against the C++ virtual machine specified in the standard. This virtual machine does not contain a stack or heap.
Generally your comment strikes me as assuming that UB is some kind of error. In practice UB is more a promise the programmer made to never do certain things, allowing the compiler to assume that these things never happen.
How UB manifests is undefined. A program that has more than zero UB cannot be assumed to be memory safe because we can't make any general assumptions about its behavior because. UB is not specified to be localized it can manifest in any way, rendering all assumptions about the program moot. In practice when focusing on specific compilers and machines we can make reasonable localized assumptions, but these are always subject to change with every new compiler version.
Memory safety is certainly critical when it comes to exploits, but even in a setting without adversaries it's absolutely crucial for reliability and portability.
> In fact Rust also silently allows signed integer overflow.
Silently for release builds, and panic in debug builds. The behavior is implementation defined and not undefined, in practice this is a subtle but crucial difference.
Take this example https://cpp.godbolt.org/z/58hnsM3Ge the only kind of UB AFAIKT is signed integer overflow, and yet we get an out-of-bounds access. If instead the behavior was implementation defined the check for overflow would not have been elided.
- I'm not sure how showing that gp can't even write a dozen lines of memory safe C proves that doing so for the exponentially harder 100+k LoC projects is feasible.
The program contains potential use of uninitialized memory UB, because scanf error return is not checked and num1 and num2 are not default initialized. And a + b can invoke signed integer overflow UB. A program with more than zero UB cannot be considered memory safe.
For example if the program runs in a context where stdin can't be read scanf will return error codes and leave the memory uninitialized.
- The thing you are looking for is convex edge geometry https://youtu.be/cZ2Y7lyzq24?t=885. Done right it can massively help with food release.
- You are right, a better wording be a "is a pure radix sort".
- Your comment made my day, thank you.
- A looong time ago I wrote my first blog post - on a now defunct website - about a VecMap where I did exactly that. Sort when needed and full flat array. That said flat_map as coined by Google is an acronym for swiss tables. See [1]. I.e. exactly what Rust's standard library `HashMap` is, also the one being tested here.
[1] https://github.com/abseil/abseil-cpp/blob/23b9b75217721040f5...
- Doing the phf as shown is an and + neg instruction and just doing % 4 is just the and. I tested it on a Apple M1 machine and saw no difference in performance at all. It's possible to go much faster with vectorization 3x on the Zen 3 machine.
> We will keep mining for these resources so where should we do it?
We don't have to keep mining, yes our lifestyle is incompatible with reducing mining output. But why is our lifestyle - or modernity in a more general sense - taken as non-negotiable? The trolley problem has a solution, stop the train.