An interesting stat I heard on Reddit and which has the ring of truth: 99% of the cost of international logistics for oceangoing trade happens in port and as a result all distances at sea are basically equivalent desirability (module time-sensitive goods, etc).
“The box” (containers and related industry) really did remake the world as we know it.
“But carbon footprint is…”
Subject to exactly the same logic. If you care about it you should care a lot about what *precisely* happens in ports.
(I sometimes think people see a really big container ship and a really small hybrid car and assume the first moving across the world contributes more carbon to e.g. a shirt purchase than the second going 10 miles, and I think “Hmm how many OOM is that wrong by. Has to be 2+?”)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Embedded finance has been a very hot topic among fintech geeks for the last few years. This week's Bits about Money goes into why I think it will be transformative for SMB banking.
I often think technologists have a poor model of banking, assuming that banks are stupid, innumerate, and incapable of employing smart technologists. All of these are very obviously false.
The bigger reason why SMB banking is mediocre is more nuanced.
It mostly comes down to "SMBs are essentially co-extensive with the sum of all human economic activity. There is no model for the sum of all human economic activity which is a good model except for the sum of all human economic activity. Now, try to write a design doc for them."
I like this sort of microintervention on the global economy delivered at scale.
Most businesses probably shouldn't have to understand how most of their customers prefer to buy things, particularly internationally. That's a minor implementation detail.
We have a lot more in the works here, including e.g. allowing you to one-click opt-in to a class of options here where accepting them is Pareto optimal (lower cost, higher conversion rate, roughly equivalent terms in expectation).
And, of course, if you're using Checkout or one of the other integration options where we're painting pixels for you, perhaps your plucky pixel painters would prefer to paint pixels precisely predicated on productively picked projects for your particular product.
In "surprisingly good experiences in government and healthcare IT", the HER-SYS system for tracking covid symptoms through self-reports also includes same symptoms input by healthcare professionals and public health workers (typically after followup call or on-site health check).
I kinda feel a little unhappy that a nationwide database with a competent web frontend that can join records by primary key is surprisingly novel, but I will take competence where I can get competence.
(Also feels surprisingly novel to read doctor's notes, which is something I've had very few experiences of in the healthcare system of either nation I've lived in. An interesting look at a norm in a state of flux.)
(It is not widely understood by retail investors that while a stock, for example, is *listed* on one stock exchange it *trades* on many separate physically distinct venues which they access simultaneously through their broker.)
Another fun example of a thing not well understood: decoupling brokerages and clearing functions is in investor interest but sometimes prevents some investors from doing something they want to do.
And since it’s obvious I do w/r/t my employer I’ll add that it was also a core design principle at Starfighter; that’s why we put everyone immediately in a live GUI that used the real API inspectable through browser.
Also why we invested effort into good docs and good error messages. (And if I did it again today I’d steal one thing Stripe does and make a first-party command line client available through Homebrew.)
A thing I would also probably do, if I were making engineering decisions again: all shipping products of company have two options for APIs:
a) /private/ namespace
b) it’s a real endpoint customers could call themselves with real docs, real support, real deprecation, etc