TacticalCoder

The individual is how he is but he's not saying anything insane. AI is not his domain of expertise yet he seems to have a pretty good grasp of how it works.

He's saying what we all know: Anthropic has no moat, OpenAI has no moat (and btw how'd that work? "AnthrOpenAIpic has a moat but nobody else does"? So it's some kind of a dualmoatopoly?).

We all saw the benchs for Google's latest models: they're beating AnthrOpenAIpic on basically every single bench.

Some of us are working with chinese models and they're now 6 months behind anymore.

So he's saying: no moat, attempt at regulatory capture to block chinese open-weight models, it doesn't mean the whole market is going to crash hard (btw a big part of the market --and already before yesterday-- had already crashed quite hard, which you don't notice if you're only in indexes for a few companies have pulled everybody up).

To this sounds... Reasonable?

What is he saying that's controversial here? When you see Google's Argon 4, are you really convinced Anthropic has a moat and an IPO valuation at 5% of the entire US GDP that's justified and that there's a possibility it'll 10x to 50% of the GDP?

Does anyone here really think Anthropic can do the kind of multiplier on its IPO that we saw with Amazon, Meta, Google, etc.?

Is it really that controversial to say that, for an IPO, Anthropic is overvalued?

Here's my bet: in a year Micron (ticker MU) went 10x in value in a year. 24x in value since after the 2022 tech flash crash. My bet is that Anthropic (ticker ANT) won't 10x in a year. And won't 24x in 4 years.

> If you access a web application, you end up re-downloading the application from the server thousands of times, and you are fully trusting the server each time. If the server gets hacked even one of the thousands of times that you open the site, you might be screwed. Regular old-fashioned applications (where you explicitly click "download") are better than this: you only need to trust the server to not be malicious or hacked at that exact moment once.

That's precisely why something like signed distributions packages are way more secure than curl bash'ing.

But many simply don't understand that. We had plenty of forged Web certificates likely. And even without that, with curl bash'ing anything sensitive you have to cross fingers that the server ain't serving one in 10 000 request with a backdoored version.

With curl bash'ing everybody is trusting every single time that the server ain't compromised or intentionally malicious.

At this point I've lost any hope: people simply don't understand security.

And as long as we'll have people saying that curl bash'ing is just as safe as signed distros packages, we'll have daily posts about major security incidents.

Security is a mindset and most people simply don't have it.

> I really resent pictures of human bodies being presented to me.

It is a reality however. Just as humans were doing human sacrifices by torturing little kids for as long as possible before killing them.

We have to keep that in mind to shield ourselves from those explaining that "all cultures are equal", "it's all relative", "it depends on your point of view", "all humans are beautiful snowflakes", etc.

Just as a reminder: Israel killed 50 000 people in Gaza... In response to people killing 1200 civilians and bringing hundreds back to their country as hostages, while the entire country was acclaiming the murderers and celebrating their deeds (killing mostly young people who were dancing at a music festival, enjoying life).

In January of this year the iranian regime sent its islamist guards to kill 30 000 peaceful, unarmed, protestors and publicly hung athletes who spoke against the islamic regime.

This is the world we live in and this is what we're up against.

I'd rather see the truth than become soft.

Which was predicted by some economists from day one.

One economist wrote, before the first Euro even circulated: "The euro will lead to too many public servants in France, too many industries in Germany, too many vacation houses in Spain [1], the default of Spain on its public debt and then the default of Greece on its public debt".

When Greece, not Spain, defaulted on its public debt, that same economist: "I was only wrong about the order of defaults of Greece and Spain".

The Euro, he predicted, would be one of the shortest lived currency in the history of currencies.

His name is Charles Gave and he's gotta be 80 y/o old now. Of course people are going to hate on him and spill all their hatred on him because he's neither a communism nor a socialist.

And then idiots are going to say that "euro is great because I can travel and pay in the same currency everywhere". As if they're own little convenience warranted the disaster that we're heading into because ideologues, not economists, decided to force a currency down the throat of most european citizens.

[1] btw there are many protests in Spain now because "costs of rents are too high" (yeah, no shit: there's a gigantic number of empty houses and apartment in Spain belonging to foreigners, who only come a few weeks per year).

> What is up with France?

I kept posting about it here over the years but at officially 57% of the GDP being public spending (and the actual percentage is bigger than that because a great many of the "private" companies that are monopolies are basically owned by state apparatchiks), France is already a planned economy.

And there's only one outcome for planned economies and its misery.

And when socialists succeed in establishing misery, they've guaranteed themselves an endless stream of voters. And they can only drive the country more into the ground. So more misery. So more socialism. So more planned economy.

France "offered" fiber optic to the entire country (at taxpayers' expense) and now that people have big Internet pipes, they're going to implement Bill Gates' old dream [1]: tax every single packet of information.

What's up with France is socialism. That's what up with France.

[1] in a leaked memo when the Internet began to take off, Bill Gates explained he wanted to tax every single network packet (for those who remember).

> I wonder though how a sensible architecture would look like.

Any withdrawal should only ever be authorized from wallets protected by multisig and every withdrawal should go through several machines, running different implementations of the exchanges' verification part of the codebase, replaying the user's entire deposit/trading/withdrawal history and verifying if the withdrawal the user's asking matches his balance/history.

Devs working on "verification before withdrawal" on one of the stack should be forbidden from ever touching the other stack, and vice-versa.

This would prevent SNAFUs like this one for now at least two devs would need to be independently owned before such a heist could be pulled off.

I'm not saying hacks would be impossible. But the "I still $6 bn in 20 minutes" would be much harder to pull.

Another sensible thing to do would be to not have $6 bn on a warm wallet: I really don't think Binance really that much at all times available on warm wallets?

Nothing is 100% secure: but the "one dev gets its laptop owned, hence $6 bn disappear" is just madness and, as a whole (not speaking about crypto here), we can really do better here.

For a start we can do better than thinking: "Nothing can ever be 100% secure, so there's no way to do better than seeing $6 bn vaporize in thin air when a single dev's laptop gets pwned" (I'm know it's not what you're saying: you're asking what can be done better).

So many are going to think: "But if you do that, it still can be hacked because this and that". Which ain't the point: the point is to make hacks harder to pull off, not to make them impossible to pull off.

> The nice thing is that using the private key requires physical touch.

Yeah (not talking about crypto): yubikeys (or similar) are really sweet for that. I've use them for ssh / pushing / fetching from Git etc. I use them for 2FA to various services etc.

There are also similar devices where you enter a PIN and the PIN to register to a service the first time and then the PIN to auth are different. So you can't be tricked by mistake to register to a new service when you just want to auth (you could still be tricked to register to a service which ain't the one you think while registering, but never while auth'ing).

And yup I use that with AI all the time: my agents go inside a container or inside a VM (or both), and the damage it can do is very limited.

It's also possible to do passthrough of devices (GPU, USB, ...) to VMs: good stuff.

More like Mt. Gox. FTX was downright fraud, from day one. SBF was already lying about Alameda's supposedly insane returns when it was all a ponzi.

He got tried and jailed for a long time for fraud.

Mt. Gox was just incompetence, like Binance here. But then on the other hand Binance is making so much money that to them $3.5 billion to cover the losses is just a slap on the wrist.

> And crypto is facilitating that.

I don't know about that. What I know is that dictators and criminals have existed since forever and the official estimates put the part of the worldwide GDP tied to criminal activities to between 3% and 5%, which is gigantic.

I don't think Bitcoin is to be blamed for that. If we're worried about dictators and criminals, a good question to ask would be: why do those exists at all? (and that's irrelevant to Bitcoin)

> That seems valuable, but I’m not sure how much truth there is to a story I read on the internet.

Wait you mean Venezuela didn't experience four years of hyper-inflation in a row? You mean you don't believe that story is true?

> IPv4 is old and broken

IPv4 is old, very old. But how is it broken? It's literally been working just fine since it came out.

That there are limited addresses is something else: it doesn't mean it's broken.

I don't care that spying/surveillance devices have more trouble spying on me with IPv4 than with IPv6 and I don't care that some apps have a harder time establishing connection.

IPv4 works. SSH over IPv4 works. Browsing the Web over IPv4 works.

Not a moat and not big but... for SMEs "Google Workspace" is a godsend from outter space.

This is what allow little SMEs to not have to suffer that piece of shit that Windows is. I know people running their SMEs (and I helped them set up Google Workspace) from Mac and Chromebooks since years and years and who do not ever want to here again from anything Microsoft.

They're basically doing everything inside "Google Workspace" (formerly "G Suite", formerly "Google apps for domain" because Google's product naming is just confusing).

Gemini is also IMO excellent for everything that is not coding. It's really rare that I use Anthropic models (which I pay a subscription for) for anything else than coding.

GMail / Drive / Google Docs / Google Spreadsheets are great for a shitload of SMEs out there who want to give Windows and Office the middle finger.

> Its sad watching them try to keep up

"try to keep up" ?

Benchmarks are just benchmarks and they could have been carefully selected but it's not "trying to keep up": they're ahead of Fable 5.1 and GTP 6 Astra in 15 of the 17 benchmarks. And in many cases by a huge margin.

Instead of talking about that, let's talk about how difficult it is to pick the model or how crazy Google's product naming methodology is.

Not saying they're not concerns: but let's not downplay what the model is achieving by saying they're "trying to pick up" because switching model ain't as easy as with the competition.

I mean: I don't use a model based on how easy it is to switch to another model. That's not what I'm looking for.

> And also, be aware that these printers require some serious development work (you need solvent to wash away uncured resin (if you ask me, this basically necessitates a chemical hood...

As a 3D printer enthusiast (I model stuff in OpenSCAD and print to fix / DIY stuff around the house/car etc.) and seen how easy 3D printing using filament as become, resin 3D printers just seem... Extremely messy to me.

I understand that for people doing stuff like miniatures resin might be more precise (but filament printers with tiny nozzles became extremely precise too) but it just feels like a lot of work compared to... Just 3D print with filament.

I may be wrong though.

> Is anybody here consciously trying to keep their Emacs configs/setup "lean", minimalist, as small as possible?

How small? I try for things to not get out of control and as I ofc version my config, I semi-regulary remove stuff I don't use anymore from my config.

But I take about 3 000 of custom elisp, written over the years, doesn't qualify as lean and minimal?

Now I gotta say: I saw packages making it into the main branch and that simplified my configuration. For example eglot: I just adopted it.

At this rate soon it's going to be very suspicious if someone is not part of any leak ; )

I'll point out that Tailscale has been repeatedly praised here as a replacement "because SSH was too old" by "security professionals". Those very same security professional happen to be the exact same closing their eyes when there's the state doing wrongdoings.

My question to those who kept praising Tailscale and making fun of good old plain SSH: how do you like your state backdoor --excuse me, I mean bug-- in Tailscale now?

> The old airs were legendary.

You gotta be kidding me. The old M1 Air had the "bendgate" where several people, me included, saw their screen die overnight for no reason. Tens and tens of pages of people complaining about it only to be with "you're holding it wrong" answers. Insanity.

I think Apple fixed that mess with the gen after that one.

> Some examples include models of Eve from Stellar Blade, Mei from Overwatch, and a Horse cock dildo.

I don't know what's more wrong: that a dumb model started downloading "Horse cock dildo" 3D models (to which end? to stuff that virtual 3D "horse cock dildo" in its own virtual LLM ass?) ...

Or that there exist "horse cock dildo" 3D models.

Seriously in which world are we living?

And I should be thinking about the people who designed that horse cock dildo and care about them? And those viewing those/using those? 3D printing those maybe? And I should care about the retards who can't even prevent their crawler from downloading 145 GB full of horse cock 3D dildo models and the like?

Nah. Fuck that, I care about my family and you're all on your own.

> And yes, there were Linux distros like Debian where people could get real computing done[0].

The author then kids in [0] that the most popular software in Debian was editors to edit X86Config but...

I fully agree that Windows was an horrible crashing piece of crap back then. Linux was a godsend to me (installed it in the early Slackware days).

Armed with Emacs and LaTeX, on Debian 1.1 then 1.2, I wrote a 440 pages book (published by an editor).

Linux was already rock stable solid, while Windows was a crashing PoS.

Linux was so solid that I'd also run it on my laptop (and back then all laptops were crappy). When I was late I'd let LaTeX convert the files to .dvi (IIRC) and the PostScript while driving to my editor. We'd then the PostScript file to the printer at my editor.

It was something else then Windows.

As for Windows, yes, I was using it too.

And the absolute best software I ever got back then was Ghost, way before Norton bought them (it then became Norton Ghost). That allowed me to take a full dump of the Windows partition and then restore it at will from that partition, when Windows would invariably become unusable.

When I discovered Linux, suddenly I had a stable system that didn't need to be re-installed every other day. There's a reason the world is powered by Linux and companies like Google weren't built on Windows.

And I also had a small consulting gig (for little SMEs) where I basically had a business installing brand new Windows systems for the clients and all their stuff (like Outlook) and then immediately imaging the entire disk.

Because the only guarantee was that Windows would be borken rather sooner than later.

TFA is right and Microsoft has always produce mediocre shit.

It's hard to say if the BSOD'ing Windows of yore was even shittier than the ad-infested, surveillance, behemoth Windows of today. Both are equally bad IMO.

Because the day Microsoft shall product a product that doesn't suck is the day they'll start producing vacuum cleaners.

> If Microsoft Windows and Office can have a valuation in the trillions then Claude eventually working its way to the same ballpark isn’t a crazy assumption.

Yeah I agree with that. However a "valuation in the trillions" ain't very different from their IPO price.

Seems like lots of risk for proportionally very little return.

A $2T valuation may not be that crazy but thinking it'll make returns like Amazon/Google/Meta/Nvidia isn't reasonable.

> There isn’t a mathematical reason Anthropic can’t be worth that in 2056.

"There isn't a mathematical reason Anthropic can't be worth 165x the actual US GDP by 2056".

Nvidia at $150B however wasn't representing anywhere near 5% of the US GDP.

If ANT lists at 5% of the US GDP and goes up 35x, it'd be worth 175% of the current US GDP.

Except in some hyperinflation scenario, I don't see how that makes any sense.

Proxmox is convenient, I run it at home. However it feels hacky: it's a bunch of scripts put together on top of Debian (with ZFS pre-installed) and I'm not betting the house on it.

I try to rely on Proxmox as little as possible and only care about the VMs. Ideally I should be able to, seamlessly, replace Proxmox with FreeBSD and bhyve (that's the plan at least).

So Tesco used to run VMs on VMWare and now moved the VMs to Proxmox: it's kinda the point of VMs. They should be easy to run somewhere else. If you've got any kind of "vendor lock-in" to your hypervisor, you're doing it wrong.

Re Proxmox: it's basically Perl scripts plus a UI for noobs (like me) in JavaScript. It's all "click click click / drop-down menu / click".

What could possibly go wrong here?

It simply doesn't feel like a serious enough project.

But in any case I don't care much: automated provisioning of my VMs is what I care about, no matter the hypervisor.

> If this is mostly competition enduced, I have to assume either their usage has plummeted or they will implode financially in short order.

If we assume it's due to competition, then their usage has precisely not plummeted because they cut prices by 80%.

> That was Motif but not CDE.

Indeed it's not CDE (although GP could just be saying that he liked old school WM / UIs like CDE, not necessarily that the one in TFV was CDE) but I think technically in the vid it's even SGI's 4Dwm. Heavily based on Motif but not totally visually identical to Motif? According to Wikipedia 4Dwm began in 1991.

As a sidenote a friend back then was lucky enough to have a SGI at home (for his stepfather was at the head of SGI Benelux) and that 4Dwm environment always struck me as particularly slick.

> he's testing thoroughly enough that he's even uncovering real bugs in Postgres, such as this numerical bug that can cause valid SP-GiST operations to fail, this format cache bug that can cause incorrect results from JSONPath, this pgcrypto bug that can cause later operations to fail, multiple memory safety bugs in the dictionary loader2, this money overflow bug, etc.

I'll keep saying it: we're eventually moving to a n-modular redundancy, with several implementations, running on different stacks/languages, cross-checking each others for bugs.

Just re-writing in Rust isn't that useful: we all move to pgrust and then what? People abandon the old software and now pgrust has a major bug and nobody detects it.

And I don't think we're only moving to n-modular redundancy at the software/application level: we'll also be moving to n-modular redundancy at the network level. Where every single packet on the subnetwork (the one where the n systems are living) has to be accounted for and identical.

And any network packet emitted by one of the implementation that won't match the others shall be suspicious and cause for an inquiry.

This shall require changes in how we build and network our stuff, but I do profoundly think that having a "majority voter" with a tiny, minimal, attack surface that's on one end exposed to the Internet and, on the other end, on a specific network (maybe with a new network protocol: or maybe with something based on UDP: not an issue with lost packets btw for it's going to a redundant system anyway) where every single network packet is accounted is going to exist at some point.

I'm dead serious: the value in pgrust is not pgrust in itself. It's that, combined with past a implementation in another language, both are now stronger.

I really hope the clue stick hits us and hits us hard with that one.

Apparently you need a 10 K+ karma account to say the following (which another account, a new one and not mine btw, posted). Some I'm just reusing the other dude's words:

In contrast, thought-terminating clichés like "facebook boomer" show critical thinking on your part.

I fully agree with that: calling someone a "facebook boomer" is just plain pathetic.

(btw the boomers are my parents' generation: don't go thinking I take it personally. I'm a GenX'er and GenX'er are badass, which is why I don't give a flying fuck saying what I think).

Something something about the pot and calling some kettle black etc.

Nice CTF! I hope if it's captured by a LLM we'll get a write up as to how it did it.

Note that Europe is quickly approaching obesity numbers the US has: child obesity in the US and in Europe are basically the same (and in both case extremely concerning).

Overweight or obesity: approx 3 out of every 4 adults in the US but already 50% to 60% of all adults in the EU (depending on the region).

I forgot the exact statistics but there's something like 90% of kids that are overweight shall stay overweight for life: so Europe is in trouble too.