cduzz

Oracle makes their money by promising the C Suite "give us money and we'll solve your problems." There aren't actually many companies in a position to credibly make that promise.

The oracle database is also actually pretty nice, according to the people I've talked to who use it.

But mostly they play in the 'nobody ever got fired for suggesting' club. These days AWS is the most egregious landlord of that club, but whatever.

I think the "party line" is that the goal shouldn't be "1950s grid, but bigger and with more red meat!"

Renewables actually cover a lot of the newer demand and can be put around where it doesn't require as much transmission capacity because the production is happening closer to the demand.

And frankly, these jumbo sized data centers can just go figure out how to power themselves. If they want to put PV and wind and some peaker plants on their campus and they can actually meet the environmental requirements, go to it. So far they seem to be dumping AI dioxin into the local waterways, but it is probably possible to run these things without grid power.

In my whole career I've seen computers getting smaller and more efficient. Sometimes you'd need more of them, especially if your product is growing, but almost never would you need these radical increases in power draw to keep up with the red queen.

In the past couple decades, the vast majority of electricity demands have gone down due to modern substitutions for things people want being way more efficient. People use LED / CFL bulbs instead of incandescent bulbs, heat pumps instead of resistive heating for water heaters and house heaters, etc.

People have also deployed lots of solar to their houses.

So by every normal measure, just by looking around outside and evaluating how I live my life, even with an electric car, my power demands have gone way down.

So the fact that there's some gooner class stroking AI and crypto coins out their network ports and making my electricity more expensive, well, yeah, I'd say that nonsense is lots of externalities that should be better managed.

What about running that same codebase on open source versions of mumps such as GT.M or Yottadb?

How do you think your code would do in GT.M or Yottadb? I understand both are open source (vs Cache).

Sorry; one final edit:

What you're describing is "bank" where the financial organizations settle transactions using a private media ("the crypto"); but each individual is still dealing with "a bank". In "crypto" you trust "the math" and you don't need to worry about the counterparty absconding with the money. In your scenario, the Argentine service provider is in a position where they trust "the app" not "the math"

Banking is fine; I'm okay with banking. Crypto is also fine, if you're doing one of the 3-4 things "crypto" is good at (mostly crime, but whatever).

But what you're describing is "bank, with settlement not using SWIFT"

Anyhow; my original statement is:

Which shitcoin or shitcoin network is "cryptocurrency" ? Do you mean you're just pushing fiat through stable coins into an exchange? Is the exchange even converting from one cryptothing to another? Are you just asking the peer to shovel bits into a some shitcoin address? If you're just putting fiat into an "exchange" and then pulling fiat out, congratulations, you've invented a pretty ordinary financial service.

You're not describing an actual financial exchange; you're making up some vague situation.

You're probably also breaking some laws around paying taxes unless someone's going out of their way to actually report these activities.

Regardless, please be specific about "pay via cryptocurrency" otherwise you may as well say "pay via paypal" or "pay via putting dollars into a bank of america account I keep open since I was a student abroad."

"cryptocurrency" is doing a lot of work here. I don't see where there's some magic "crypto ledger" in your getting paid for a service. Just an internet service that promises to transmit money from one place to another.

I think I offered a pretty concrete example of a real situation and how the imagined utility of crypto magic isn't actually all that helpful.

Now, perhaps what you're suggesting is "I've got millions of units of money that I've acquired through locally illegal means and I want to transport that money someplace else and I don't feel like renting a private airplane and diamonds up my prison wallet are uncomfortable and I don't want to pay taxes or otherwise explain why I have a brand new S class I paid cash for." then yes, I'll accept that there is some grey area where crypto has a place (to explicitly work around legal regimes I dislike, would like to degrade further, and don't feel like dealing with).

Can you describe a "society collapse" where your cold wallet _can_ get you out of trouble, and why that particular situation isn't subject to "I will shoot your corgi." coercion?

If you're dealing with a peer you can't trust (like the US government) that will ask for your passphrase and then not torture you for it if you refuse, I'd suggest you're not really in a situation where society has broken down.

Let's talk;

You're in Crimea right now. You want some gasoline. You have some magic numbers in your magic rock. It needs you to cast a spell to transmit those magic numbers to someone else. I guess it probably also needs ... some electricity and some network access and some time to make those magic numbers update in the gas station's magic number.

In what way is this actually going to help you in a total collapse? Let's stand around over my iphone and cold wallet and spacex terminal while we wait for things to beep the right number of beeps?

In the off-chance you find yourself in this situation, what's to stop the party with the gasoline from just pointing a gun at you or your corgi and asking you to add some zeros to the number of magic beans you're zapping into their magic number?

Now, perhaps you as an international agent of mystery, can use some shiny rocks or magic pieces of paper to make your escape, and you can then, once safely back in a stable society, use your magic numbers and spells to transfer "wealth" to some organization closer to your new physical location (let's say belize?) -- in which case, to avoid a bunch of tedious laws and such, perhaps that magic trick will work in the way you suggest.

But, when things go haywire and there's just Hobb's all against all, I'm not sure a magic cold wallet of certificates is going to do much.

People have made crazy stories up about "the origin of money" for a long time; Adam Smith makes up a zany story about people agreeing to use chunks of metal to make change for bartering because it's too hard to keep chickens in your wallet when you need to break a sheep[1]...

The notion that people gravitate to metal vs "fractional sheep" is without historical evidence. People in fact, a long time ago, did trade things like IOUs or pieces of broken reeds or sticks[2]. Those sticks acted like IOUs and were often traded around far past the original parties to the contract.

Money is a contract between you, your peer/counterparty, and whatever organization you "trust" to mediate if someone really isn't happy with the outcome of a situation.

Money is, indeed, culture.

[1] https://www.adamsmithworks.org/documents/chapter-iv-of-the-o... [2] https://en.wikipedia.org/wiki/Tally_stick

I feel like "open source" in this context is, as you say, an uncomfortable nuance; the tooling (llama.cpp, et al) is open but useless without weights.

The weights are extraordinarily expensive "capital" that is donated by big organizations who are all at war with each other.

I don't know that it will ever be possible for, for instance, archive.org, to make truly open weights. And, other than archive.org, I can't imagine any other "open source" organization (freebsd? apache?) being in any position at all to make truly open weights.

Maybe governments, government organizations, or universities.

None of whom are currently funded, mandated, inclined, or particularly interested in dumping the money into buying the infrastructure needed to make weights.

The localhost troll works better if you use the decimal representation of it:

http://2130706433

or any integer multiple of that 2130706433

These days I think "at night" is mostly covered or at least could be mostly covered both by wind and batteries.

The "base load" question may still be appropriate for deep winter, high (or low) latitudes, etc, but renewables are getting there pretty fast.

AI data centers that exist and are operational are running at maximum capacity. That's why you see things like the tiny little data center run by xai showing up as a valuable resource to xai (on the sale side) and anthropic (buy side). It is "only" 300 megawatts and there's a 1.25 billion rent on it per month.

If all these other data centers were anywhere near coming on line, that 300mw data center would be a rounding error not a line item as it is right now.

So someone's signed contracts for way more and way larger data centers, someone's purchased billions in hardware for these not yet operational data centers. I'm wondering how depreciation's going to work on all these assets...

Anyhow, I'm not really sure what "max capacity" is here, nor am I really aware when they're going to be delivering the operational assets that are currently levered to their eyeballs and consuming 1/3rd of the memory made on the planet.

As far as inference vs training, have new gotten radically better than old models or only marginally (at the cost of 10x or more the training costs)?

Very exciting stuff.

This is actually what happens.

I run my word processing software on my apple 2 (a total joke of a computer) instead of running it on the WANG.

I run my book keeping software on visicalc instead of the IBM.

I run my simulation software on my IBM PC (I even paid for the 8087!) instead of the VAX.

Moore's law has, at least so far, allowed the pioneers with toy computers to grow their toys big enough to solve "big boy" problems after some time has allowed the toy computers to be faster and the pioneers have scaled their crappy home-grown solution to solve their 60% of the problem that was originally solved by some enormous complex system.

Eventually the toy infrastructure gets expensive and solves 90-120% of the "big iron" problem space, but it also grows to cost as much as the big iron solution, but then a new generation of toy software and toy systems emerges to disrupt the "big iron" systems.

See also http://www.catb.org/jargon/html/W/wheel-of-reincarnation.htm...

High Bandwidth Memory uses thousands of interconnects for the data bus. DDR style memory typically uses in the neighborhood of 64 bit transfers at a time.

HBM tends to be integrated onto the package (board, multi chip module, die) because there are really tight signaling and wire routing constraints that make "modularity" impossible.

I remember back in the day you could get motherboards for your 286, 386, and sometimes even 486 with external L1 / L2 / L3 cache -- you'd buy a bunch of static ram dips that you'd populate sockets next to the CPU, and set a bios or DIP switch to enable it. These days that's just not practical because there are too many wires interconnecting the cache to the dies and cache coherence logic, and the speed of light is just too slow and electricity is too messy to put "external" to the die/chip/package, even if the packaging issues could be addressed.

HBM memory is similar -- it's not practical to make a generic interconnect that'd actually work reliably enough to provide field replaceable memory modules as you can with DDR style dimms.

EDIT:

Apparently I'm totally wrong in that these "SOCAMM2" modules have thousands of pads (like a CPU socket) and can in fact run with the same data bus width (1024 bits wide!) as "local" HBM. Very cool. And please ignore my out of date blatherings above. It's still not quite as fast as if you put the HBM in the package, but it's way faster than the DDR style setup.

They've got thunderbolt 5 which is pretty good and every port has a dedicated controller. You can even network macs together just over thunderbolt.

Not as fast as raw PCIe slots inside the chassis, but I don't think the ghost of steve jobs cares much. He'd tell you that if you've got no taste at all you can put his beautiful machine into ugly junk like from sonnettech and you can go do ugly things elsewhere.

I think apple's happy taking the market they've got and they'll leave the big guns HPC market to nvidia. The margins look great for nvidia right now, but I suspect nvidia's path will be similar to dram's boom/bust cycle more than apple's continuous "premium tool" brand's market positioning.

That 1tb / 12 channels is a continuous streaming read / write rates? I assume big wide DDR memory, for random "IO", is much slower than compared to HBM.

I feel like at a certain point there are just going to be big SOC packages with 128gb of ram and stacks of cores (each with their own "local" cache) and the 128gb "local" HBM on-package ram will just be the 4th or 5th level cache, and big server boards will have 4 of those and CXL elsewhere for "main" memory.

And things like the VAST stuff also blur lines between high speed local storage and less performant san or bulk commodity storage.

The old memory / storage hierarchies are getting mixed up (again).

Interesting times.

NEW VISTA, OUTER RIM—Just a cycle ago, the brain was in a living person. Now, hours after its first owner died, it sits on a slab draped in tubes that quiver as they pump liters of blood substitute and other fluids through the organ, supplying oxygen and removing waste. As far as anyone knows, with many of its key functions intact but maybe awarness muffled by drugs, the brain hovers between life and death. As people subject it to experimental drugs, sensors record the brain's reactions, capturing hundreds of data points on its cells, proteins, and physiology. Then, after 24 hours in this state, it will be sliced into hundreds of pieces for more detailed study.

Ah; so it _is_ a tool problem. It is _also_ a moderation problem.

One could ban orgs that flood the zone with AI generated trash, but is there some potential middle ground where there are sets of filters to identify duplicated bugs, and possibly just internally dump "AI spam" to a lower queue?

This seems like the sort of problem I'd addressed in the 90s with killfiles and spamassassin. In other words, can't the ingestion just go through some filters to shield the humans at the end of the pipe?

If the AI is awesome at identifying security bugs in the linux kernel, it likely can also identify if the thing it's found is similar to something that is already found in the security mailing list?

Or, put another way -- what flags the duplicate? The filer or the system? If my cheese factory is measured by the volume of cheese instead of the quality, I'll churn out the cheese even if it's sloppy duplicated cheese. And that is the case if a person has to flag a new ticket as "same as this" or not.

What's that law that says that any sufficiently large problem turns into a moderation problem?

Hydrogen is no good for cars because it sneaks into the metal and ruins it (hydrogen embrittlement). Your car is mostly metal, especially the part that stores the fuel and the part that makes it go boom.

You can probably make electricity directly from H2, and you can probably make special pressure vessels that'll store that H2 (though even then it'll have a 7 year "inspect thoroughly" and a 15 year "throw it out regardless" lifespan.)

H2 is a silly fuel unless you're making rockets. Or if you're trying to distract people.

I'm going to mix my metaphors a bit here...

There's the saying "Any idiot can build a bridge; it takes an engineer to build a bridge that barely stands."

To put this another way, any idiotic LLM can write code. It takes a person with domain experience to understand what code to write, rewrite, or not write.

I've seen lots of organizations hollow out their internal competence in favor of outsourcing the skills. LLMs are the ultimate expression of that. There are people who say "you need to have people in your organization who understand how things work because they're the ones who solve problems!" and there are other people who say "focus on your core competencies! These problems you're worried about aren't your core competencies, so get rid of those experts, they're expensive and annoying; we can just sign a contract with an organization that'll know things for us."

At some point we all will identify exactly how much "seed corn" you need for the next season. We'll figure that out because we're starving, but at least we'll all know.

I think this is mixing domains quite a bit;

If I'm talking to a friend or peer and I'm on a crappy link, we can probably work it out. If I'm calling my lawyer from prison with my "one call" I really want my lawyer to get my instructions clearly and correctly, ideally the first time without a lot of coaching.

Where on this scale does "person talking to LLM" fit?

I believe there's a ton of research into the shannon limit and human speech. You can trivially observe how much redundancy there is by listening to a podcast at 1x, 1.2x, 1.5x, 2x, etc, and when you can't follow what's going on, you've found the "redundancy" built into that language. This number falls way off when you're listening to a person with an accent or when the recording is noisy or whatever.

You'll also find that your tolerance for lossy media is radically different based on latency and echos and jitter in the audio (which I believe is the point of the original "don't use webrtc" article...)

Finally, people may tolerate this, but the "phonem to token" thinger may be less tolerant, and will certainly not be able to magic correct meaning from lost packets, and if the resulting exchange is extremely expensive or important (from the lawyer and the "I'm in jail in poughkeepsie; I need bail!" exchange) you really want to take the time to get it right, not make things guess.

Well, with a car without lane keeping or cruise control, it'll try to kill you pretty quickly if you stop actively controlling the car....

With an AP1 car, you can scan ahead 5-10 seconds (in about 1 second) and pretty quickly assess if the car's going to have a hard time with anything coming up (mostly it is an issue of lane lines vanishing in highway curves or lanes splitting ambiguously).

It is in the happy medium of predictably stupid such that it isn't ever really trusted. Something smarter may lull you into trusting it more, which leads to situations where it can trick you... I imagine there are people who think it's safe enough to scan the road every 1-2 minutes when in a more automated vehicle, and obviously the ultimate goal is "people in car ignore the driving aspect of the trip" -- both of those seem pretty ambitious goals but people are spending gigatons of money to solve these, so maybe it'll be solved?

I've also had animals jump in front of me (just in general, not related to teslas); driving is just dangerous but is a lot more convenient than walking everywhere.

There are those who are paranoid and those who are expedient. If you're truly paranoid, you spin up the thing you want to run, measure what it does, and open the holes to allow it to do what it needs to. It's tedious and sometimes error-prone, but in some environments it is necessary.

In the vast majority of the world, you set permissions to what's reasonable and trust that most of the time things will work out pretty well and have a plan for if you need to fix things on the fly.

I personally am not terribly paranoid, but I've worked places where we had to be pretty paranoid (shared hosting).

I'd have guessed that the default paranoia-first policy would be "drop everything; verify what you need" which would include AF_ALG.

share and enjoy!

I've been quite happy with my "first generation" tesla with the mobileye system. It has only tried to kill me a couple times in 6 years of driving it; it is not terribly smart but within the system's limits it is very stable. I certainly don't trust it to drive unattended, but it does offload 5-30% of the toil of driving on highways, which is pretty nice. Offloading 50-80% but constantly wondering "is it going to try to kill me?" I don't think would be as relaxing, though I understand lots of people have chosen to just not worry, which I guess is fine...

At the time I got the car I wasn't sure if I wanted the old "totally obsolete" AP1 or the "probably going to get way better (cough)" AP2; I'm glad I got the obsolete version....

I wonder if there are modern cars with systems comparable to the mobileye system from the original tesla setup.

Venus seems like a wonderful place to live, relatively speaking.

At the right altitude where you can "float" on the ocean, it's a pretty comfortable temperature and there's plenty of solar energy but you're shielded from the solar radiation. So, long term, your body will still work, assuming you can solve "the other problems."

Of course, the down-side is that there's nothing to stand on and probably more importantly, there aren't many useful materials to work with besides tons of carbon, oxygen, and nitrogen. Not much hydrogen there, so not much water, which probably is the biggest problem. One of them, anyhow. Also, there's probably not a whole lot to do besides float (zoom, actually) around and slowly go stir crazy in your bubble.

But relatively speaking, it's way nicer than living in a hole on mars where you'll slowly die from gravity sickness, or radiation poisoning, or whatever.

I suspect this is one of those "it depends" situations; does the 128gb vs 64gb sku have more chips or denser chips? If "more chips" probably it'll draw a tiny bit more power than the smaller version. If the "denser" chips, it may be "more power draw" but such a tiny difference that it's immaterial.

Similarly, having more cache may mean less SSD activity, which may mean less energy draw overall.

If I had a chip to put on the roulette table of this "what if" I'd put it on the "it won't make a difference in the real world in any meaningful way" square.