A few things should be made clear.

1. Buying in January was almost the best case scenario, it is not a good metric to generalize from.
2. This is not remotely as safe an investment as buying tech stocks. It is much more all or nothing. Especially people new to this should understand they should not spend more then they are willing to have completely vanish even if they are correct in direction, and stock go up, just they don't go up enough fast enough.
3. Options can be complicated, and you can losing substantially more than just the initial purchase price if you don't know what you are doing, or even if you do but are not actively managing your portfolio. One scenario is the options are in the money and you hold through execution. In the period between the execution and when you are first able to sell at open, the stock can drop in value below the execution price.

$ϵ$-random is a bad baseline because activation space is not isotropic (or some other reason I do not understand) and this is not actually that unexpected or interesting.

Isn't this just the answer? To rephrase:

The SAE is only able to represent a subset of the possible directions from the initial space when you force it to compress the space down.

If you take a magnitude from a direction where change matters, and then apply the magnitude to random dimensions most of which the model throws away, it will result in a smaller change.

Grainger is the first thing that came to mind as a legitimate reference that has pricing online.

I don't think you have invalidated any of my points, that the hp of a tesla motor doesn't make any sense to compare with hp ratings of an excavator motor, that high hp electric powerplants are very expensive, that adding VFD's which are not operationally necessary is a large cost and is most often left off in favor of running the motors.

Diesel-electric operate on electric where available, and switch to diesel when they get to unelectrified areas. Many cities are electrified within maybe a 20 mile radius of city center, with farther branches being diesel.

Tesla motors would not be rated at a tesla vehicles rated hp for an industrial application, they are not expected to operate at 100% duty cycle.

If you want to look at industrial motors for sale at a gold standard place like Grainger, a 250hp (sticker hp Model 3 equivalent) motor is $30,000 - $40,000 just for the motor, no VFD, no mounting, no installation.

https://www.grainger.com/category/motors/ac-motors/definite-purpose-motors?attrs=HP%7C250&filters=attrs

Another $15,000 - $30,000 for the VFD

https://www.grainger.com/search/motors/motor-drives-speed-controls/variable-frequency-drives-accessories/variable-frequency-drives?attrs=Maximum+Output+Power%7C250+hp&filters=attrs&searchQuery=VFD&sst=4&tv_optin=true

Just the power cable that you need for a 400A 3 phase service might run something like $40-50/foot on the floor? It is so expensive you can't even get a good reference on Google. Have to hope you have 3 phase available at all, and that you are close to the pole. A housing site with an excavator you might want 300'? 500'? Another $12,000 - $20,000

So far it is at $67,000 - $90,000 just sitting on the floor, and you need to be highly certified to touch any of those things in a commercial settings.

Edit: Just to add, the above motor weights 3000lb and uses something like 130kw/hour, so even a Tesla 3 battery would last approximately 30 minutes.

While lots of hobbyists use VSDs on their equipment so they can run 3 phase motors off standard US line voltage (I do this for my 1/2hp CNC spindle, 1 1/2hp bandsaw in the past), these are mostly non-certified and small Chinese VSDs, that require programming and knowledge beyond the scope of what even skilled operators are normally capable of, yet alone allowed to do in electrical regulations.

The reality is a certified piece of equipment, professionally installed, can cost in the $500/horsepower range. On even a fairly modest piece of equipment with say 5x 3hp motors, that starts to become quite large percent of the cost of the equipment tacked on, and a huge amount of added complexity. The cost of these drives does not scale linearly with size.

The only motors I have used in a professional setting with a VSD required very precise speed control (spindles), not heavy equipment.

Just FYI for many cases, heavy electric motors you leave running same as a diesel. The inrush on them is massive and is often charged separately as peak 'demand', the costs of which can dwarf the pure kilowatt hour charges of running the motor (the kind which most residential users are used to).

Switching large currents also will wear the components, they have limited a cycle life, so leads to expensive replacement and downtime.

Places I have worked would chew you out if you turned something off that was coming back on within an hour.

Large motors will also likely be 3 phase, which greatly limits locations they can be used.

I am trying to make a general statement about models and contexts, and thinking about the consequences of applying the concept to AI.

Another example could be Newtonian versus Relativistic physics. There is a trade off of something like efficiency/simplicity/interpretability versus accuracy/precision/complexity. Both models have contexts in which they are the more valid model to use. If you try to force both models to exist at once, you lose both sets of advantages. You will cause and amplify errors if you try to interchange them arbitrarily.

So we don't combine the two, but instead try to understand when and why we should chose to adopt one model over the other.

Group coordination evolved because it was vital for survival at a time when the group size was not the entire population. There was a margin for error. If some percent of groups took themselves out due to bad models, the entire species wasn't eliminated. A different group that also coordinated eventual filled the empty position.

Since the human group size is now global in many aspects, the damage from a bad model could result in elimination of the species. The expected value functions are changed. The process above can still play out, but there is a very real chance the group that eventual takes over as a result of their better model is not human, because there are none left. We instead want the outcome for the species tied to a better model than it is now. It does not seem like that is a possibility if there is insistence on the human model maintaining control.

I definitely understand this perspective and do agree. Less Wrong was not the only target audience. I tried to start in a style that could be easily understood by any reader to the point of them seeing enough value to try to get through the whole thing, while gradually increasing the level of discourse, and reducing the falsifiable arguments. I tried to take a broad approach to introduce the context I am working from, as I have much more to say and needed someplace to try to start building a bridge. I also just wanted to see if anyone cared enough for me to keep investing time. I have a much larger more academic style work which this is pretty much the abstract of, that I have spent many weeks on over the last year working on, and I just can't seem to finish. There are too many branches. Personally I prefer a list of bullet points, but it becomes very easy to dismiss an entire document. Here is how I might summarize the points above.

• Human brains are flawed by design, and generally cannot see this flaw, because of the flaw.
• Humans should be trying to eliminate the impact of this flaw.
• The gap between flawed models and working models, sometime can not be closed, and instead requires a totally new model.
• AI is a totally new model that does not need this flaw, but insisting on converging the models reintroduces the flaw.