Longreads
Forbes profiles Peak6 and Apex Clearing; Peak6 started out in derivatives trading, ran into trouble with their nearly-insolvent clearing company, and ended up acquiring it in order to offer clearing services to other fast-growth companies. In finance, there seems to be a tendency for the companies that are good at engineering to be more agile about moving to parts of the supply chain where they have a bigger competitive advantage. When investing is a systematic discipline, the skillset is more translatable to other things adjacent to it.
Rohit at Strange Loop Canon writes about how some small teams can quickly learn to build a nuclear weapon, while larger and better-funded ones can forget how to. Especially interesting is the riff on poaching talent but not being able to use it, and on the low ROI of industrial espionage. This implies that a distinct company culture can be a particularly effective defensive tool; the people who thrive in a particular company culture are less likely to succeed anywhere else.
Estimating the cost of technical debt: this paper argues that states whose unemployment benefit systems were written in COBOL caused delays in unemployment benefits with an aggregate GDP impact of $215bn in current dollars. (I have written before on the case for either subsidizing or banning COBOL.) (Via Marginal Revolution.)
Casey Handmer on why Starship is a big deal. If going to space was in some sense worth it at the old cost structure, how much more worthwhile is it when the cost drops by orders of magnitude?
Scholars Stage has a great meditation on the two Bush presidencies: the GHWB administration had numerous foreign policy triumphs, GWB had many failures, and the people in charge were often exactly the same.
Books
Vaclav Smil, Natural Gas: Fuel for the 21st Century: Peter Zeihan has gotten a lot of mileage out of analyzing economics and geopolitics by paying attention to actual geography—as it turns out, soil, rainfall, mountain ranges, and natural resource deposits are a big deal, and ignoring them is perilous. Vaclav Smil has a sort of twist on this: the book is about the history of, economics of, and utility of natural gas, and a recurring theme is that Smil defaults to thinking in terms of chemical reactions. Different mental models come in and out of favor over time (Zeihan-style thinking was pretty unfashionable in the 90s, for example), and my guess is that the chemistry-fluent thinkers have a good decade ahead of them as more and more decisions start with Fermi estimates of emissions impacts.
Open Thread
As always, share any links that Diff readers would enjoy.
In light of yesterday's riff on Meta, let's talk about the Metaverse: who has a head start (besides Roblox), who is investing the most (besides Meta), who gets the most spillover benefits (besides Nvidia—the demand for triangles that refresh 60 times a second will keep on exploding)?
A Word From Our Sponsors
Bitcoin and Ethereum have been on a tear recently, and if history shows us anything, it's that altcoins will start to get attention next.
Crypto Pragmatist is an email newsletter that gives you the in-depth information you need to understand what's happening in altcoins, and what's coming next.
Subscribe to their free weekly newsletter for data-backed crypto investing research, or check out their paid subscription for bi-monthly in-depth altcoin reports.
The makers of big game engines (Unity, Epic) and related tooling will see large spillover effects from the investments into the Metaverse. (I'm long U)
The people who make Star Citizen have some very good FOIP tech. Uses what’s already in most webcams to kinda accurately map an avatar’s facial expressions to the user’s facial expressions. Not sure how that’ll work with the big vr goggles, though.
From an access standpoint, you’d want as many people in the metaverse as possible but the necessary peripherals are still pricey and niche. Almost everyone has a smartphone and most modern phones can do some duty as a vr headset (google cardboard is a good example). If you want this thing to scale, you can’t expect every user will eventually have a clumsy set of goggles and controllers. Phones seem like a great bridge because they are already ubiquitous, especially if they can get rid of the controllers entirely and go fully gesture-based. Apple and Samsung seem well placed now, but in 2003, you’d have said Nokia and Sony were well placed for the next big thing. Amazon’s phone had all those sensors…
High speed cheap broadband, preferably wireless/cellular is necessary and gets more necessary if the processing power is off of the user’s device. Since gaming was mentioned, even very intense online games use little bandwidth compared to, say, streaming video or sending files. This is because all the heavy lifting is done locally on the CPU/GPU. You don’t send the whole image through the internet at HD (or greater) quality at 60fps (or greater) to be processed externally and then sent back. It’s all in house. That has to change for the metaverse to be truly omnipresent and not just something you do at your desk. You’ll kill your device’s battery in a few hours, or faster, if you are running even moderately demanding 3D graphics like those demoed for Meta. More cpu/gpu means more power is needed, more heat is generated by the device, and means you need a larger device. Or, you take all that work and put it in the cloud (see, for example, Google Stadia or Steam’s in home streaming for a local version). This seems like the model that will win out.
That means fast, deep, reliable, widespread broadband internet for whatever device you need. When my friend demoed Horizon Workrooms for me, he couldn’t get his wifi signal strong enough to reach his living room sofa where we were putting on his pair of oculus headsets. Obviously some of that is building design, materials, layout, etc. but these constraints are real and are barriers to a truly omnipresent metaverse like we see in sci-fi. I get one bar of cellular signal in my apartment and no 5G signal.
So, we might ask, who is best positioned to provide the *infrastructure* for the metaverse even down the the last mile, to the thing that is in every potential user’s hands?