I like these guys but get annoyed at the 1kWH == 1 Typical House. While a case can be made that over a 24h period a house may consume 24kWh the fact is that "typical" houses burst to anywhere between 5kW and 10kW depending on appliances (AC + Freezer compressor + Oven baking a roast and boom you're at 10kW) As a result if you hook a thousand houses to a 1MW reactor you're going to get a crap ton of brownouts.

<raises hand> Two EVs, electric oven, refrigerator, air conditioning here... We installed solar when we moved in but would need massive storage to handle peak loads "locally" at our house.

Burst loads are the wrong number too. You still want to average over many households, but not average over wide swaths of time. Maybe you need to design for a peak of 75% of AC units plus 15% of ovens plus 20% of fridges being on simultaneously.

True, it is more complex than it might seem at first glance. One of my EE classes had us go out to a power plant to tour it and they talked about some of the complexity (because grid wide you get larger swings). They used to count on something like 80% of the households turning on the oven at 5PM while dinner was being prepared. They had actuarial staff that really dug into these numbers because some conditions would require starting up a 'peaker' plant and that took 3 - 4 hours to bring online from dead stop, about 1.5 hours from a "warm" start.

I imagine the temperature probably doesn't vary enough for a 1k household size area to allow for only 75% turning on the AC when it hits 32C

There are duty cycles to take into account.