I want to be part of sick edits like this

This is not sustainable. I predict instead DCs will build their own cheaper, more reliable power plants, utilities will then purchase access to them on terms favorable to the DC operators. Why? The net present economic value of the marginal MWh is far far higher for AI training and inference than default electricity consumption patterns. This observation seems controversial but it's self evidently true.

when even a single noble soul understands something I say in the way that I want to say it I feel calm

Peter is the original creator of Pokemon Red RL that beat the first gym. Go check out his latest project! Very cool alife

We achieved gold medal level performance on this year's IMO! Our model thinks and writes proofs in clear, plain‑English - no formal code required. Unlike the narrower systems used in past competitions, our model is built to reason broadly, far beyond contest problems.

Same thing happened to me this past 6 months. So painfully obvious. Turned my follow up question into something that I knew would trip an LLM up. It did. Had a good laugh about it. Rejected the candidate

The bar for AGI is that it can program a harness for me to use it to write software with my voice

So basically you're free to say anything you're not forbidden to say. Right.

Foiled by magic smoke btw

Been deep in an eVTOL aeromechanics rabbit hole. I was originally most interested in Lift+Cruise (L+C) for a two-seat design because it seemed mechanically simpler. But after reviewing lessons from teams at Aurora, Beta, Wisk and Joby the hidden complexity has become very real. The core challenge is severe vibration during transition to forward flight. Oscillating prop loads can hit ~3x the steady, average load. And because we rely on differential thrust and hence variable prop RPM for control, the rotors often sweep through frequencies that resonate with structures. This becomes a massive driver of stress cycles and structural sizing. As one very seasoned eVTOL engineering leader said about scaling prototypes: "aero gets easier but structures get harder". The deeper issue is the prediction gap. Small-scale tests frequently miss critical problems like the boom resonance one of the companies above discovered. Even when simulation flags it, simpler predictive models can severely underestimate the actual loads. One company even used stick-on accelerometers to identify vibration paths and issues in testing. This makes carefully instrumented, full-scale (or near-full-scale) testing is essential if you want truly reliable data that can be evaluated against predictions. So the "simple" L+C path demands mastering the repeated stress of starting and stopping lift props every flight. Tiltrotors have more mechanical complexity, but they at least confine their toughest aerodynamics to a shorter time frame when the rotors are near perpendicular to the ground. They also have their own problems, but are less likely to surprise late in the development cycle. Hybrids gain some mechanical savings (fewer tilt axes) but still face both edgewise and parking vibration problems. Look at how Archer had to switch to four blade lifting props, which has significant drag and range penalty but reduces the vibratory loads vs two blade props. Ultimately, it's clear the real moat in AAM isn’t just building a flying prototype—it's the validated mastery of these dynamic loads to ensure long-term safety and reliability. Huge respect to the teams publishing their data.