narnarpapadaddy

That one notch further on the spectrum that’s being described. Will “clear large chunks of dirt in the field” be found to have damaged information? Is there genuinely nothing to learn from the chunks of dirt? Or is it still too expensive to pay attention?

Most undergrad CS students go through an algorithms course and understand a bubble sort is inherently slower than a quick sort, no amount of optimization will get you past the fundamental n^2 time complexity. You can generalize those same analysis principles to whole architectures and distributed systems. If you don’t recognize that the server interactions your solution requires boils down to an n^2 algorithm, you’re in for a bad time. Unlike a function a live, deployed system is typically harder to evolve. Sometimes constraints like backwards compatibility are such that you simply can’t fix it.

“Batch is the primitive”

Yeah, but what if we are at the point we don’t need machinists any more?

The rich get richer, even if you’re enriched solely with an above-average intelligence. The bar for break even just moved up. If the US thought 13 million unemployed/underemployed people was problem, just wait until it’s 30 million. They could afford a 10% drag on the economy, who wants to bet they can double it to 20%?

That relationship is (was?) a proxy for two mature superpowers. The dynamic between the mid-tier powers in the Middle East is not that similar, IMO. With the US realigning, I expect to see more activity in that conflict.

I can’t know the future, but if the model I’m proposing is correct we’ll see some NK/SK shake-ups soon. That won’t bode well for Iran.

It makes an invasion of Iran less likely, but it also probably means more conflict, not less.

Acquiring nukes rebalances the power between Iran and the US and its neighbors. Roughly equal pairs of states fight more than lopsided ones, historically. The probability for a win on either side is higher when on equal footing.

Nukes as a deterrent caps escalation. This emboldens the holder of the nukes in lower levels of conflict. See Russia with Ukraine (knowing NATO wouldn’t respond), Kargil after Pakistan got nukes, and NK sinking the Cheonan.

Iran getting a bomb probably triggers Saudi Arabia to follow suite, possibly Turkey and Egypt too. Many of the conflicts in the area aren’t (solely) defined by US presence. The Iran–Saudi, Sunni–Shia, and Israel–Iran rivalries to name a few.

It’s possible the dynamics shake out differently, but color me skeptical.

For humans hallucinations are a particular class of error, so I find hallucination more descriptive than either error or bug.

I also think it’s relevant because a hallucinator often doesn’t recognize that the hallucination isn’t real. That’s more accurate for the LLM than either lie or confabulation, IMO. They algorithm is trained to produce strings of text that have semantic meaning based on some statistical likelihood of tokens appearing next to each other. The LLM algorithm is working as intended.

Hallucinations are also often emergent from a particular state or situation, which reflects the generative aspect of LLMs.

Hallucinations are sometimes resolved in humans by grounding exercises. “Touching grass.” The same is true for LLM hallucinations. Inaccuracies are found by cross-checking the output against an internet search or another LLM.

No disagreement here; I felt countless smaller problems leading up to the final blow was implied.

I think this is the correct take. And until we have an AI Hiroshima it’ll be difficult to get the international community to work together. It’ll take rogue AI (or AI-powered group) disabling a significant world power before everyone comes to the table. Otherwise, it just looks like MAD and the equilibrium holding to the powers that be.

The “elements for randomness” are carve-outs for the parts of territory the map can’t contain. The more granular the map, the larger relative proportion that falls into that set. Nobody can predict the layout of my basement from a globe of the world.

We know with a high degree of certainty the universe contains things we can’t predict or observe, see Bell’s Theorem.

“Truth” doesn’t exist outside reality. There’s no substrate to hang it in. Information theory is likewise a subset of the territory, part of the universe, not apart from it. For these things to exist independently, you need something other than or bigger than the universe to put them in. If such a thing existed, sure, from that perspective maybe a lossless compression could exist. But that’s a metaphysical argument.

The currently accepted answer to that is “the territory appears to be randomized” (quantum mechanics)

Trust as delegation is an interesting angle; if you’ve not read Team Topologies I’d recommend it. These topics are treated pretty well.

> is advocating for and what doesn't tend to happen in dysfunctional organizations.

Agreed the observations are interesting, and that’s what is being advocated for, I just don’t think top-down vs bottom-up is the axis that impacts that as much as disturbed vs… I said “centralized” but probably “localized” is a better term.

An aspect that’s missing from this analysis is distributed vs centralized decision authority. In militaries and in tech startups a “squad” or “two-pizza team” are empowered to make decisions for their local domain because they have the real-time context. The decentralized decision authority that happens with matrix-style management is a much higher contributor to coordination overhead than the top-down or bottom-up axis described here.

An autonomous team that does what they are told or an autonomous team is free to pursue opportunities as they see fit are both useful and can react quickly. The question is how much scaled alignment you need to execute a particular goal.

You can’t invade a country with one squad of 10,000 soldiers or 1,000 squads of 10 that do as they please. You need to delegate execution authority to small teams who are aligned to a shared goal.

Similarly, you can’t expect one group 10,000 devs with no internal structure to explore and find interesting opportunities to explore; you’ll just get noise. Nor can you try to manage 10,000 devs directly. You won’t be able to grok the context.

Complexity management in organizations follows the same patterns as in code. Single responsibility, don’t repeat yourself too much, encapsulation, etc.

It’s really a function of information theory, the more people you have to ramp up or keep current on context, the lower your ability to make and stick to decisions.

So the wiggle room for efficiency is in the mechanism for counteracting yaw? Either the weight or energy expense of running a tail rotor or quadrotors or variable pitch mounts (or some other clever mechanism)?