boring-human

Indeed: https://en.wikipedia.org/wiki/Dual-sector_model

Knock-off tokens from a replica router, that's hilarious. "If you look closely, the logo says Clod."

It should have been "eight reasons to use std::simd". Inefficient.

If you're afraid of buying derivatives from a trustworthy party that may have contractual obligations with Anthropic, but them from a trustworthy party that doesn't. If they expected a $100 post-IPO price one year from now, they'd short-sell you a derivative for $200 or whatever. Derivatives are legal, and as ethical as anything else in finance (more ethical than just about all crypto.)

If you trust the other party, you can always write a derivative contract. If the other party was long, their position is closed. If they weren't, they're short now.

This would have been better as a "spoiler alert."

Vacuum-tube amplifiers are not in the same class with techniques that are unlikely to have any perceptible influence on what you hear.

I don't disagree. Not in the same class, but the audience overlaps.

one could make a modern amplifier reproduce any quirky behavior of vacuum tubes, e.g. a higher and frequency-variable impedance or certain kinds of distortions, but usually nobody bothers to do this, because it would be expensive

In other words, the willingness to pay for OG tube amplifiers exceeds the willingness to pay for the sound thereof. I'm not sure you disagree with me either.

Not to mention the cost of the gold-plated audiophile cables to wire it...

The true argument is about quantity - of people, not code. All qualitative arguments are missing the point.

Cut me some slack, Jack.

Even after we've all retired (pretty soon for those who can afford it) or transitioned out of software engineering (for those who can't), we'll still get to amuse each other with home-brew projects like this. Warm fuzzy feeling - I'll take it!

I harbor some hope that the (sad) fall of human SWEs will at least be accompanied by language defragmentation. We don't need 38 systems languages once human taste is mostly out of the picture.

Did you upgrade the tool binaries? I also couldn't see it until after the upgrade.

I don't disagree with your findings, but here's the model I use:

- Fiber: ^

- Dairy: v

- Coffee: ^^

- NSAIDs: vv

- Ice cream splurges: vvv

I don't have any anecdotal data, just detecting a whiff of a possible pattern in your statement. DDoS is bots. Any chance the prevalent discourse is bots? "I ain't saying she a gold digger..."

We just don't trust them enough to not have human pilots.

Much of the value of a human crew is as an implicit dogfooding warranty for the passengers. If it wasn't safe to fly, the pilots wouldn't risk it day after day.

To think of it, it'd be nice if they posted anonymized third-party psych evaluations of the cockpit crew on the wall by the restrooms. The cabin crew would probably appreciate that too.

I don't disagree. However, if we change nothing, one likely alternative outcome is planet of 100 trillionaires, 10000 concubines, and an ever-shrinking ghetto of scavenging paupers. The solution is to turn the un-meritocratic nature of this particular bit of technical change against it.

As the value of labor plummets, more GDP will accrue to capital. But to whose, exactly? Let's categorize individual investing performance as a function of luck, corruption and skill. Only skill is meritocratic, and there is no good reason to reward the other two. Things have trended away from skill in recent decades.

As AI automation progresses, it provides more of the skill. Eventually all investing decisions will be AI-based, democratizing the process but effectively leaving luck and corruption in control of who wins.

At that point, there's just no good reason to reward individual investment performance. Since luck averages out, corruption will largely determine who the 100 trillionaires are.

The solution is to tax away the portion of investment returns that are not based on skill, which will trend towards 100%.

Yep. I think the idea that the benchmark is determinative is just as deluded as the notion that it should be unbreakable.

Benchmarks are on the honor system. Even the tightest benchmark can be cheated. If the benchmark is so secret and air-gapped that it can't be cheated by models, it can be cheated by its own authors. You can't use benchmarks to gate out cheating.

If you don't have the honor system in mind when you're reading scores, you're wasting your time. Is it some unknown outfit with wild claims? Is it connected to Epstein, Russia, the real estate "industry", or sleazeballing in general? Do they have previous history of ratgaming the numbers? Replace its scores with asterisks and move on.

This is an odd choice of a thread for a laundry list of complaints about AI and about a person that, say what you will, is nowhere near the list of planetary "really bad guys". Even if we limit it to tech, the list starts with someone way richer, then goes through four or five way-shadier people.

If you're OK with victim-shaming here, doesn't it say more about you than Altman? What does it say about your viewpoint?

I understand the temptation to Streisand this, but for the love of, please don't. S1/2 were the best show I've seen on TV. It would be a crime against good taste.

He hasn't kept ahead of the destruction of the dollar very well.

You can't price dollars in gold to measure value. Gold doesn't measure value better than the dollar at any point in time, let alone over time. Just use the price index for one currency, or the relative price indexes across currencies.

Suits in agriculture don't drive the combine either, a farmer does. The other 99% of pre-automation farmers went on to other jobs. They happened to be better jobs than farming, but that's not necessarily always the case.

could we say it’s beyond AGI at least in cybersecurity?

AGI is like the Holy Grail. Either in the Arthurian Hero's Journey sense, or in the sense of having been a myth all along.

Same background as you, and same exact experience as you. Opus and Gemini have not come close to Codex for C++ work. I also run exclusively on xhigh. Its handling of complexity is unmatched.

At least until next week when Mythos and GPT 6 throw it all up in the air again.

Yep, I think the lede might be buried here and we're probably cooked (assuming you mean SWEs, but the writing has been on the wall for 4 months.)

I guess I'm still excited. What's my new profession going to be? Longer term, are we going to solve diseases and aging? Or are the ranks going to thin from 10B to 10000 trillionaires and world-scale con-artist misanthropes plus their concubines?

There's a joke in here somewhere, I just don't know where.

I kind of think that these threads are destined to fossilize quickly. Most every syllogism about LLMs from 2024 looks quaint now.

A more interesting question is whether there's really a future for running a coding agent on a non-highest setting. I haven't seen anything near "Shall I implement it? No" in quite a while.

Unless perhaps the highest-tier accounts go from $200 to $20K/mo.

I still haven't found one to equal Railroad Tycoon 3, which has this kind of neat reactive diffusion field pricing engine.

My feelings exactly. I still play it regularly for this reason alone. A really amazing game under the hood.

The Rhodes Unfinished scenario is probably my favorite. On the highest-difficulty level, it starts out as a hard map, then becomes an insatiable resource grab. By the end you're building vanity suspension bridges over chasms and digging tunnels the width of Kilimanjaro.

I think AI "slop" will improve medical diagnoses dramatically. Let's assume for a second that the first specialist did not graduate at the top of their class.

The year is 2030, when LLMs are more pervasive. The first specialist now asks you to wait, heads into the other room and double-checks their ET diagnosis with AI. Doing so has become standard practice to avoid malpractice suits. The model persuades them to diagnose PV, avoiding a Type-II error.

But let's say the model gets it wrong too. You eventually visit the second specialist, who did graduate at the top of their class. The model says ET, but the specialist is smart enough to tell that the model is wrong. There is some risk that the second specialist takes the CYA route, but I'd expect them not to. They diagnose PV, avoiding a Type-I error.

Love this. I guess there was no application for "Me not that kind of Orc"? That's the best one, but maybe it has hints of homophobia.