Also, fun fact: the campus tree atlas says this tree was planted in the 1960s. This adds to my impression that Storrs is generally bad at growing big trees.

setting aside the question of whether this would all be worthwhile if it was going to work, I think observations like this should really make people question whether the fundamental premise that unimaginable computational scale is the key to AGI is correct.

seems appropriate as tuesday is where my unrealistic assumptions about how the rest of the week is going to go peak.

I don't know about triple, but I have seen what I'm convinced was quadruple, where the reviewers, the authors, the editors and the subject matter are all anonymous.

I think this and your "almost intuitive" comment are the crux of the issue.

the difference is 2.4% for a kibi- vs kilo-, but almost 10% for tebi- vs tera-! and 12.6% for pebi- vs peta-!

xkcd.com/394/

but it's so easy!

1 GiB = 1073741824 bytes
1 TiB = 1024 * 1073741824 = 1099511627776 bytes!

What could be more intuitive!

It all seems like such a huge fraud to me.

If these guys are so good, why shouldn't they get peanuts for their base salaries and then massive bonuses for performance?

Their good runs never last because 1) they are based on luck and 2) to the extent they had some secret sauce, it gets figured out quickly and neutralized because things are hypercompetitive.

They all honestly seem more like celebrity mutual fund managers than anything else. They beat the market for a few years, everyone thinks they're magic, and then they regress to the mean.

But I also have a genuine question: conditional on spending obscene amounts on football coaching, what justifies these kinds of contracts? I'm no expert, but it seems to me that *every* college football coach (except maybe Nick Saban) is basically just a flash in the pan.