I'm a co-founder of Nootrobox. We're super excited to really invest into the rigor and quantitative metrics behind cognitive modulation. I'll be floating around answering any questions if you folks have them.
A combination of things - a super old/competitive market (processing), the ever changing risk dynamics and Square's targeting
Let me illustrate points 1 and 2 with an example.
a) Square earns $2 in revenue per $100 of transaction volume
b) $0.5 would be their gross margin accounting for expenses i.e. bank/visa fees
c) The next biggest line item to subtract would be risk. If there is a fraudulent transaction of $100, then square has to compensate it with 200 transactions of $100 each. (200 txns *0.5 gross margin). Net effect is not only do they have to deal with low margins, they also have to manage risk very well. And this is not something you accomplish overnight especially as you enter new markets.
Personal anecdote: I had $120 or so disappear from my Starbucks wallet a couple of months ago and the merchant had to eat the loss.
Lastly, large payment processors get out of this loop because they manage the big (Targets and Walmarts of the world) with the small and over time have fine tuned their risk engine. Square went into it in the reverse order, targeting mom-and-pop stores first (which have low volumes and high acquisition costs) and then trying their hand at large merchants (Startbucks in this case and they lost a ton doing this). In the process they never got risk management right for either segment.
interesting. we use coffeescript at nootrobox.com, but as you noted, it doesn't have a lot of hand rails. many a time, i've run javascript through a js2coffee translator to verify. will look into typescript.
Was pretty surprising when we first saw it, but have seen since that lots of sites now serve different prices/coupons depending on the device you use to access.
Yes, I encourage you to do your own research. A lot of the chemicals have not been formally studied on a significant group of healthy adults. A lot of the models in studies are mice models or alzheimer patients, etc.