I feel like even a human would also fail if given all data results but those for x and then get tested on x just as the function results differ from previously observed behavior.
Is it not more interesting to observe how the model(human or not) incorporates the new data to better match reality in the case of distribution shift or other irregular distributions and others
Yeah I'd believe a stronger sedentary lifestyle is the biggest factor in the waste line stagnation if the diet is not the same.
They did not have omnipresent smartphones in 1970.
Another 10% was low % 5yrs bonds which they had to sell early with penalties to cover the run on the bank which incurred considerable losses.
I would think a VC bank knows that some of its VC investments would not outperform.
My understanding is that if the run does not happen, it's business as usual, nothing like '08 when borrowers started failing to pay en mass on internationally overrated loans in AAApackages.
There are still plenty of opportunities out there. Layoffs are a social contagion that do very little, if anything, for a company’s financial health. This is all about the trend and trying to “discipline” labor. But it won’t work, especially this time because the US doesn’t have decent childcare (many mother won’t return to the workforce for a year or more) and has blocked millions of immigrants who would normally have entered the country the past several years.
That wouldn’t be a problem for engineers who are good at networking or the interview song and dance, but plenty of talented engineers aren’t particularly good at those things.
The company will lose some talent, sure, but let’s not pretend this isn’t a strategic time for a layoff or pay cut.
