3 ms·
The problem with hacking the bell curve like this, is that the bell curve will continually adjust itself within the new parameters. There will always be 'fring
by BLKNSLVR 7y ago
The problem with hacking the bell curve like this, is that the bell curve will continually adjust itself within the new parameters. There will always be 'fringes' until the sample space of "things being measured" is too small for the bell curve to apply.
In this case for Amazon, it's reaching a limit of capitalism: only selling the most common and profitable items in order to maximize profit and minimize waste (which itself maximizes profit). This will have the side-effect of making the "things they no longer sell" have greater value due to their scarcity, and in turn greater profitability as a 'boutique' product.
Similar theory to the unrelated topic of coal versus renewable energy, in that as renewables gain popularity the demand for coal will go down, and therefore the price goes down as well, lowering the point at which coal and renewables compete for being the cheaper source of energy (carbon price notwithstanding).
Another example, although far more theoretical, is that of personality profiling. What are the fringes? Once you've removed the "terrorist"-level fringes, does the system re-adjust to a new normal and give a new set of fringes to target?
Forced normalization as an end-result of pure data analytics (AI, ML whatever you want to call it).
I'm assuming these effects are covered by fairly basic economic theory (which I haven't studied, but somehow find more and more interesting as I get older).