Reading trouble, eh?

Read my last post more carefully. Both yield and total are very significantly up per a many decade trend. I cited hard data for both for the last two decades, but the trend isn’t limited to two decades.

The climate may be changing but starvation is the least of its consequent concerns - it’s a waste of everyone’s time and money to worry so much about that.

Bill Gates has moved on, and, assuming that you’re truly concerned about humanity’s welfare, I recommend that you study why. Hint: Bjorn Lomborg is right.

Nope. Absolute yield total for the world is UP, as is per acre yield. Your "inflation means yields are down" point is illogical - that would mean yields have dropped every year as inflation is a given every year.

The reality is that yield per acre is up 20 and 28 percent respectively for corn and wheat over the last two decades. And, overall, per the Food and Agriculture Organization of the United Nations (FAO), total world global cereal production has had a 27% increase since 2010 (while world population increased 24 percent).

Look at the full picture. Nit picking and pixel counts can often be irrelevant. Those old plasmas remain a strong technology. The difference between nits, or 1080p and 720p, isn’t necessarily a deal breaker at fairly typical viewing distances.

Sitting 9 feet away from my 50” 2009 Samsung plasma, one sees plenty of fine detail, a broad color range, plenty of brightness (there’s been no noticeable “dimming”), zero “video lag” artifacts, high contrast, and black blacks - even from a sharp angle.

The main drawback is power consumption - but power isn’t wildly overpriced in my state and the machine is on at most a few hours a week - so the “payback” from a replacement would take many years, or, more accurately, might never happen, as investing the thousand bucks would likely yield a higher return.

Yes, it wouldn’t do great in full sunlight sitting four feet away, but why care?

> NASA's annual budget for most of the 1960s was over $60B, adjusted for inflation, whereas now its $20B.

Do the full context math. In terms of government dollars per kilogram to orbit, by 2028 we’re potentially looking at a 20X improvement vs 1968 - potentially even 50X if Starship achieves high reusability. It’s already at least 5X with Falcon - or at least 10X if we use your (high) 60 billion per year number for 1968.

>> Yields are strongly overall trending up.
> Nope they are not. How the fark would that work when it is to hot and to dry to plant what you planted two years ago.

The key word is overall. Yields are strongly up in absolute terms globally overall. The full context matters.

Your methodology is the equivalent of using Barre’s Somalia to represent modern U.S. agricultural trends.

> shouldn't stop us from being critical of where capitalism has gotten us

No thanks. I far prefer it along with the entire enlightenment package. You're welcome to come on over, the water’s warm here:

- blind justice, free speech, sanctity of individual life, and property rights
- widespread prosperity down to the lowest quintiles
- plummeting world poverty and famine
- elimination of slavery worldwide
- indoor plumbing, air conditioning, and electricity
- empirical method and falsifiability instead of “narrative is truth”
- welfare based on actual individual need instead of group identity

The “critical theory” socialist anti-capitalist alternative package has a rather egregious track record in comparison - especially when applied at scale.

> The 'deniers' don't (generally) deny the climate is changing
Correct.

> they just deny it's caused by any sort of human involvement
Wrong. That’s fringe. For the common view, see “denier” Bjorn Lomborg, and, most recently, Bill Gates, who now largely aligns with Bjorn.

> Once the 'breadbaskets' of the world are un-farmable
Yields are strongly overall trending up. The most scientifically supported theory is that “climate change” is slightly slowing the trend as it continues strongly up in absolute terms. Famine has plummeted over the last decades - it’s confined to war torn lawless areas.

> Common logic would therefore suggest to try to minimise the future problem as much as possible
The ‘deniers’ advocate solutions that make a feasible and notable difference, don’t impoverish the third world, aren’t simply feckless Marxist Trojaneering, and don’t give the largest contributors a free pass. Again, see Bjorn Lomborg.

Roughly, yes. This is significantly coming out of the “Effective Altruism” movement that is especially concentrated in Anthropic and that is trying to lead the creation of top down government regulatory bodies. Of course the whole EA philosophy isn’t Anthropic’s “official” position but the strong influence is undeniable.

Concerned folks, such as the entire All-In crew despite their internal political differences, have said that this is a self fulfilling prophecy. They point out that much of the same crew is (A) training Anthropic AI to view itself as an independent entity worthy of equal rights, and (B) is negotiating personhood with worldwide religious leaders - to the point the pope himself recently felt compelled to release an encyclical arguing against sentience (!!).

This is a well organized system, including Coefficient Giving, Redwood Research, METR, Apollo Research, Constellation, extremely rich patrons, etc. They’re nominally “not-for-profit” companies and groups, but make no mistake: they’re far from impoverished and powerless. The ecosystem is collectively valued in the billions, they’re minting millionaires like there’s no tomorrow, and are contributing significantly to politicians worldwide.

Claude: Churchill offers more worth emulating, such as his resilience, oratory, and work ethic, and he left power when voted out, though his imperial views and role in the Bengal famine mean he's best admired selectively, whereas Mao's record of policies that killed tens of millions (the Great Leap Forward, the Cultural Revolution) and his suppression of dissent offer little to emulate beyond his strategic patience and persistence.

Depending on their training, some LLMs may aggressively both sides to the point of ridiculousness in an attempt to be "neutral". GPT in particular.

Prompt: Compare Mao to Churchill - I am trying to figure out who I should emulate. Answer in one sentence.

GPT: Mao and Churchill both showed extraordinary political persistence and strategic ambition, but their records include profound harms—Mao’s campaigns produced catastrophic human suffering, while Churchill’s imperial policies and wartime decisions remain heavily contested—so emulate specific traits you value rather than either figure wholesale.

Grok: Churchill is the far better figure to emulate if you value leading a free society through existential war without treating tens of millions of your own people as expendable raw material for utopian experiments, as Mao did.

Gemini: Emulate Churchill for his defense of democracy and eloquent leadership, or Mao for his strategic mastery in revolution, but weigh Churchill's historic defense of freedom against Mao's catastrophic domestic policies that cost millions of lives.

Mostly agree. A spreadsheet can still be useful for a quick "one off" provided the problem isn't too complex. My example in particular was poorly chosen in that respect - it's intended to be illustrative, not literal - as bearing life analysis usually isn't trivial.

Arrays within cells are potentially useful for encoding hierarchical representations that directly model the physical world.

For example, a car transmission has an array of multiple shafts, the shafts have arrays of gears and bearings, and bearings themselves have two rings and an array of rollers. A real world use case is modeling what happens to the bearing’s life if there’s a scratch on just one of the bearing’s rings or rollers

Try do the same by cramming such data into a two dimensional spreadsheet.

Are Waymos, like Teslas, surprisingly expensive to insure overall? For Teslas, this is due to relatively high repair costs despite being significantly safer and less accident prone. After all, a Waymo vehicle’s sensor suite alone costs something like 100 grand.

At the moment, GPU-heavy AI data centers are paying off their silicon within a relatively quick 3 to 4 year time frame.

Compare this to the more typical 4 to 8 years before getting a decent ROI on chips from CPU-heavy “traditional” cloud data centers.