🧠 Fintech is disrupting Wall St. Nubank over 100m users.

Welcome to Fintech Brainfood, the weekly deep dive into Fintech news, events, and analysis. Join the 40,219 others by clicking below. 

Hey Fintech Nerds 👋

Nubank passed 100m customers, Monzo raised again, Affirm, Dave, Robinhood crushed their earnings.

Fintech has no chill.

I had to cover Nubank in Things to Know 👀 this week, because the what now question is fascinating. More markets? Which ones? Why?

For you embedded finance junkies FIS getting into BaaS is also in Things to Know 👀

My 📣 Rant this week is Capital markets x Fintech. It’s finally here. Everywhere I look, and if you’re not seeing it I’m ranting at you.

PS. I’ll be migrating from Substack to Beehiiv next week so if you wonder why things look a little different that’s it.

PPS. Keep an eye out for a special edition on Wednesday 15th. Cooking up something special 🎧🧠

Here's this week's Brainfood in summary

📣 Rant: Capital markets x Fintech

💸 4 Fintech Companies:

1. Govern GPT - The RFP Database for Asset Managers & Funds

2. Proov AI - An AI to validate AI model compliance

3. Agent Smyth - Stock Analyst AI

4. Lucra - KYC and Fraud detection for Law Firms and Funds (UK)

👀 Things to Know:

1. Nubank hits 100m users - Brazil is done, where next?

2. FIS launches a BaaS platform - Strategy makes sense but can they execute?

📚 Good Read: The CFPB Vision for a Data Economy

(Also, I haven’t bothered mentioning the Synapse/Evolve stuff because Jason has you covered better than I could)

Weekly Rant 📣

Capital markets x Fintech

The final hurdle for Fintech disruption has been and will always be capital markets. 

Very few of us have ever worked in Wall St, making the problems seem distant. The products and jargon are often so dense that you pretty much have to have worked in finance to make any sense of it.

But capital markets and financial markets are the beating heart of a well-functioning economy.

We tend to think of financial services in terms of the bits we can see and touch, and “Wall St” always seems disconnected. But the reality is Wall St is just a bank (or, more appropriately, a series of banks and funds) for bigger businesses and governments.

If you’re a shop owner wanting to go national, you might issue a bond instead of getting a loan from a bank. If you’re successful at going national, you might even go international and then list stock on the NYSE. 

Capital Markets allow

• Investment in growth, like buying new equipment or locations, leading to more products sold and jobs created

• Efficiency, because there’s competition bidding to invest in the bonds, the price can be comparatively better than a traditional lender might offer

• Risk diversifying. With many buyers and sellers, no one investor has to be over indexed on the shop owner. Indeed the shop owner can even buy bonds or stock in other companies or from the government. 

The result is money gets where it is needed, makes an economy attractive to global investors, and makes the country they operate in more prosperous. 

This attracts the best talent, investors, and global companies.

That’s why companies like ARM list on the NASDAQ, and it’s likely Klarna will too.

There’s just one problem.

Capital markets haven’t really innovated since the 1970s.

The last major technology shift was the transition from paper stock and bond certificates to computerization, a process called “dematerialization.” (I kid you not.) 

Despite some of the most sophisticated quant trading algorithms and high-frequency trading AI, everything that happens after an order is placed looks like bad spaghetti. 

I think there are three disruptive forces

1. Fintech APIs

2. Fintech AIs

3. Tokenization

They apply across all products and services, and there’s a patchwork of companies and founders attacking them.

But first, let’s explain why it’s taken so long to get here.

The complexity curve

Fintech has slowly worked its way through payments, consumer neobanks, spend management, brokerage, and now treasury management.

The real final boss is financial markets.

I’ve used this picture to describe how Fintech has taken the last decade to reach this complexity curve. As we go further up, the number of potential customers decreases, and the complexity of customer needs increases.

The market shift is that we now have a generation of founders who’ve worked for several years at Wall St. and have seen how stuck it truly is in the 1970s. These founders are building new products across the spectrum.

And financial products in capital markets are complex.

So let’s grossly over-simplify to help build a map and layer the complexity back on.

The map of products 

This is my mental model, lets step through it piece by piece.

a) Securities are most well-known and available, like stocks and bonds. 

Stocks or bonds can be bought easily via most brokerage apps or services if they are publicly listed (public) securities. They have a clear regulatory framework allowing consumers to buy them directly (AKA, become the ultimate beneficial owner, or UBO). Shares in Apple or PepsiCo, or US Treasuries can be bought this way, as well as funds like S&P 500 index ETFs. 

Private securities like credit or equity are the equivalent of stocks or bonds but do not trade on primary trading venues. This is where an auto lender might secure capital from companies like Blackstone or Apollo to have the liquidity to create loans. Private equity can be everything from a start-up that just has VCs and founders on its cap table to companies like Nuvei, which were listed and had their shares bought out by a specialist firm (Private Equity firm or PE firm) to invest in its future growth. 

b) Commodities and Alternatives are “everything else.” 

Any asset that is not a security (SEC-regulated or MIFID-regulated in Europe) is generally classified as an alternative. The most well-known of these are commodities (like oil, gold, corn, or copper). Commodities are raw materials that drive our economy and are subject to massive speculation and price spikes.  

This leaves everything else that falls into the “alternatives” bucket. This could include anything from wine, to art, to music royalties and intellectual property (IP). Generally, commodities are regulated by the Commodities Trading Futures Commission (CFTC) in the USA, and most alternatives are regulated under AFMID in Europe.

c) Derivatives are contracts that “derive” their value from underlying assets.

This one took me a while to get, but the most helpful way I’ve found to think about it is to imagine your mortgage. Let’s imagine you had a variable rate mortgage (one that goes up and down as rates change). Imagine you bought a variable today because interest rates are high, but in the future you’d like to lock-in a fixed rate if rates come down.

Now, imagine someone else will let you swap that variable rate for a fixed one when rates come down. You might feel better about the certainty of a fixed rate, especially if its low, where a professional investor can get value by speculating on the movements of rates. This is an example of a swap.

There are several main derivative types

1. Futures. A relatively simple agreement to buy an underlying asset (e.g. Oil) in the future at a set (strike) price. 

2. Options. A more complex agreement granting the right, but not the obligation to buy or sell an asset at a future date for a an agreed (strike) price. Options have an irritatingly dense language around them where the option to buy is called a call option. The option to sell is called a put option. 

3. Swaps. Interest rates and FX rates are commonly swapped to give businesses certainty of forecasting and financial planning.

4. Complex combinations of the above. A Swaption is an option to buy a swap. These agreements can become highly bespoke.

These products get complicated fast, especially when you’re dealing with future-dated purchases or sales and logic like “if this, then that.” That’s why historically, regulators and politicians have held that these products are not for consumers. 

The International Swaps and Derivatives Association (ISDA) has done good work standarizing many of these contract types but the logic (often on page 97), usually looks like software logic “if x, then do y.” 

Note: Structured products are a fun rabbit hole. Let's say you want to create an ultra-bespoke strategy for your global company, which is exposed to hundreds of currencies. In some markets, you need options, swaps, and futures, you’re issuing several bonds, and you have any number of other complex transactions. Banks and funds will help construct a structured product around these more complex needs.

The actors

a) Buy side (asset managers) vs sell side (banks). 

Historically banks have been the issuers of bonds and securities. They’d work with their corporates to diligence a bond (like your bank might with you for a loan), then they help bring it to market where pension funds, hedge funds and institutional buyers would invest in that bond. The banks sell the bond so are refered to as sell side. The funds and fund managers buy the bonds so are refered to as buy side.

This pattern plays out with IPOs too, if Goldman Sachs is bringing an IPO to market, they’ll look to sell that IPO to as many funds for a reasonable price prior to listing day. 

b) Custodian banks hold the assets on behalf of legal owners (UBO). 

A custodian bank can be thought of as the vault in a bank branch. They hold and protect the asset (e.g., a mortgage, bond, or stock) but also manage a complex lifecycle of activities around the asset (e.g., dividends being issued)

So while you see your stock at Robinhood, it’s managed on your behalf via a custodian like BNY Mellon, State Street, or Northern Trust. The stock may also physically be recorded at a depository (like the DTCC). 

Why doesn’t the brokerage do this directly? It comes down to the separation of concerns. The brokerage benefits from an information asymmetry. They know more about your assets than you do. The custodian in part, ensures this is less likely by monetizing the holding of the asset and the management of its lifecycle.

c) Financial market infrastructure (FMI). 

Financial market infrastructures (or sometimes intermediaries) play numerous roles. 

1. Exchanges. The most obvious one is the exchange, like NASDAQ, NYSE or LSE. These exchanges list securities and connect buyers and sellers (buy side and sell side). 

2. Clearing. Companies like Apex Clearing or JP Morgan manage the delivery of an asset for a payment (delivery vs payment or DvP). This sounds like a simple payment, but the reality is countless regulatory checks have to happen before this move can occur.

3. Depositories. Are where assets live, and where brokerages and funds can buy those assets by posting collateral (e.g., Robinhood might post 2% of their expected daily holdings at the DTCC instead of 100%). Centralized Securities Depositories (CSDs) are a legal requirement in most markets in the post-financial crisis world and are responsible for ensuring all market participants have enough collateral to cover their obligations.

d) Analysts and Ratings (not pictured) Analysts play a critical role in creating indexes and scoring risk. Where consumer debt might have a FICO score, bonds typically have a rating agency score (like AAA or junk). Where your personal net worth and future might be something you worry about, public stocks are covered and rated by stock rating analysts. Moody’s, S&P, and Fitch are the most well-known ratings agencies, whereas banks like Goldman and BofA would rate stocks. 

e) Distribution is the part you see (not pictured). The bit that person on the street sees is the tip of the iceberg. Your Vanguard, E*Trade, or Robinhood account is the last mile, and where orders are taken and current holdings are displayed. 

Important note: Most brokerages play multiple roles. For example, Fidelity and Vanguard are major asset managers and “buy-side” institutions in their own right.

f) Regulators have evolved over time, and jurisdiction is always national, not global. You can map regulatory frameworks over parts of the broader market map, but no single regulator owns the whole thing in the US. Often each jurisdiction has an equivalent. 

While they try to work together internationally via the G20 and the Committee on Payments and Market Infrastructure (CPMI), the reality is that the situation is a maze.

Capital Markets are:

• Highly regulated.

• Highly manual.

• Hard to upgrade.

• Full of sunk costs.

But that’s changing.

The big shift in markets 

a) Banks have played less of a role since 2008. After the financial crisis, banks were forced to hold more capital reserves and collateral. This means it became less economical to compete for some lending activities. While the supply of bank lending may have fallen, demand has not. 

One area this can be seen most clearly is in private credit which had $375 billion in assets under management globally in 2008. This ballooned to over $1.6 trillion by March 2023, and is projected to exceed $3.5 trillion by 2028. Its market share is nearing that of syndicated loans and high-yield bonds (historically bank products).

Another example is private equity. Companies are delaying their IPOs, with some opting for direct listings, SPACs, or even hitting profit and just staying private (which is low-key what I think Stripe is doing). The banks are fundamentally less competitive, and their product offerings are, in many cases, not what the market wants.

This opportunity is being taken up by buy-side actors and funds like Apollo, Blackstone, and Franklin Templeton who are more than willing to take up the slack and have often been left frustrated by their bank partners' lack of innovation and dynamism. Alternative assets under management are expected to grow from $10.7 trillion in 2020 to $14 trillion by 2025. 

b) Alternatives lack market structure like exchanges or CSDs. These markets have historically been much more peer-to-peer. Securing a line of private credit or a secondary private equity sale is a complex world that only sophisticated investors and corporations have access to. 

As a result, the upfront cost of a transaction can be high, with diligence and KYC being manual and one-off. The private funds are trying to change this by building their own marketplaces, but there isn’t an NYSE for alts. Nor are there clear rules of the road and regulations as you’d have with securities.

Yet “alts” have historically outperformed securities on a risk-adjusted basis. This is why most large funds and family offices have a ~20% allocation to alts)

c) We haven’t had a major tech upgrade since the 1970s and are due one. Alts are growing and lack market structure, securities are inefficient, and a new generation of distribution exists (e.g., Neobanks, mobile brokerage, treasury management for growth companies).

Innovation won’t initially come to the incumbent banks. It will start with smaller funds, smaller asset classes and on the edge. The story of Wall St is endless cycles of companies being founded that start small in some emerging area that later go on to become massive. 

This time, I think they’ll default to using Fintech infrastructure, AI, and tokens.

Fintech Op1: Capital Markets x Fintech APIs and distribution

This trend has been running for the past decade and is making its way across the underlying financial products spectrum (albeit not deep into derivatives yet). 

a) BaaS for markets? API-first companies are already here. Companies like Drivewealth and Atomic Invest have collectively brought securities and some alternatives to market with APIs. They’re a new form of registered broker-dealer (that, in many cases, may also do clearing). Instead of selling an asset to the end customer, they enable their clients like Neobanks or spend management platforms to integrate buying stocks, treasuries or other assets into the consumer or business's everyday dashboard. 

The natural tailwind these companies have is the US capital market. Whether its treasuries, stocks, or credit is the most liquid in the world, and its assets are in the highest demand. 

b) Platforms for debt raising. Companies like Finley, VaaS, and Setpoint help manage a debt facility with calculations, reporting, and the management of funding requests (not all of the cash hits your bank at once). The default move here would be spreadsheets, but you could bring a platform in-house. 

Alternatively, the B2B Neobanks and treasury management startups are starting to fill this hole, too. It’s uncanny how often I see BNY Mellon as the listed custodian for spend management/SPIC registered broker-dealers who are selling US treasuries to growth businesses. Arc’s* venture debt marketplace comes to mind.

c) Next Gen trading platforms and broker dealers. Younger funds and fund managers will likely adopt the lowest friction UI for trading. Architect* is an example of a company who offers derivatives trading to the pro-sumer and startup fund, or mature funds looking for an ultra high performance execution platform.

Fintech Op2: Capital Markets x AI (Analyst as an AI)

Everywhere I look, I see Capital Markets x AI. Usually, one former Wall St analyst plus one deep AI co-founder sees the opportunity to take a workflow and make it 10x or 100x more efficient. This comes in two categories.

a) AI analysts and agents. Agent Smyth analyzes stock and sector data to help traders prepare for their day or make a buy/sell decision. Lucite provides companies' business overviews, competitive analysis, and financial metrics as an analyst might export from Pitchbook. Finster is a former deepmind and JP Morgan team-building financial data analysts. GovernGPT helps manage the RFP response process for asset managers and funds. Every RFP response is ultimately delivered as a PDF or Excel and involves a lot of copying and pasting. The trick is making the language fit.

A lot of this tech is still buggy, and it's early days. A highly skilled human outperforms the default model usage (like all AI). But that’s not the point. 

Anyone who’s worked in AI and machine learning for a while will tell you the trick is continuous model improvement to zero in on a problem space. These wedge products will eventually work, and the first adopters will be a new generation of funds entering capital markets or in their early days. 

b) AI workflow tools. PDFs are the new oil. This model is popping up all over in Fintech, but it’s especially applicable to capital markets. Everything in capital markets runs on a PDF or spreadsheet, from KYB to ISDA master agreements. I’ve seen companies collecting 20+ documents for customer onboarding or underwriting. 

Imagine you’re a lender looking for a facility from Blackstone or Apollo to lend to new customers. You might have to submit 20 or 30 due diligence documents and materials. You then have to repeat this for every credit provider in the capital markets. Arc* has built a marketplace to make this more efficient and uses AI for document workflow automation.

Fintech Op 3: Capital Markets x Tokenization

a) Tokenization of money market funds allows 24/7 access. Blackrock has launched a tokenized money market fund BlackRock’s new BUIDL fund, a month after its launch, managing $304 million in assets. Franklin Templeton a little bit more recently noted it has achieved $384 million in assets (I suspect Blackrock is still ahead here). But the question is why.

It’s common for most companies to park cash in a money market fund overnight while the business isn’t operating, but then they have to wait for cut-off times at their bank and the markets to move anything. Both funds allow money to flow in Stablecoins 

b) The tokenization of all assets is next. This will come from two ends of the spectrum. Firstly, simple existing assets like cash (Stablecoins, CBDCs) and money market funds get tokenized by market incumbents in partnership with start-ups (e.g. Franklin Templeton is working with Polygon, Blackrock with Securitize, Owners, and Archax). We’ll also see natively digital assets like Bitcoin come into the market more broadly. Next will be an Ethereum ETF, but play that forward a decade.

Finally, all of these will have to be interoperable. There are countless attempts to do this from incumbents (like Canton Network) or disruptors (e.g. Polkadot or Baselayer). These things will co-exist for a while, but eventually, we’ll see the bridge between existing assets and the new world of tokens be fully built. 

(Pro tip: You’re systematically not paying enough attention to Canton Network)

Fintech x Capital Markets is here

Three things are happening.

1. The buyer is more willing to look at non-traditional vendors. When the buyer changes, so does their software. An “elderly millennial” is now 42 years old, and has grown up with the internet. They increasingly have buying power in senior positions in capital markets (and the wider economy). 

2. A new generation of funds and fund managers will adopt new tools as their default operating system. When new funds and companies are founded, their infrastructure and products tend to differ. You’ll see founders posting about their “stack,” which is usually G-Suite and Slack, but also, Gong, Webflow, and other SaaS tools. 

3. Incumbents are finally getting there with tokenization. Tokenization will make markets default 24/7 and globally, create new asset classes, and make all assets programmable. That will create space for innovation.

These three things lead to a 10-year shift that will become the biggest tech upgrade in financial markets since the 1970s. (Incidentally, the 1970s were when the “buy side” behemoths like Blackstone and Bridgewater got their start.)

The adage “this won’t work for capital markets” is incorrect. 

My buddy Eduardo said to me earlier.

We’re in phase 1. Whenever I speak to people with tenure and experience in capital markets, they tell me disruption is impossible. The market is too fragmented to be transformed, or if it is, incumbents will lead it.

But I’m getting that feeling again. 

That feeling that there are just too many little loose signals that real change is coming.

That’s exciting.

ST.

4 Fintech Companies 💸

1. Govern GPT - The RFP Database for Asset Managers & Funds

Govern creates a central knowledge database of marketing materials, old RFP responses, and product documentation. It updates itself based on any new questionnaires submitted. The goal is to take due diligence questionnaires from days to hours. 

🧠 Every questionnaire and RFP asks broadly the same stuff in different formats. Finding that from multiple sources and making it coherent is not difficult, but it is time-consuming. Every enterprise sales team and consultant wants this product. I spent many years pulling together giant RFP responses and proposals, and there was always something we did somewhere that didn't quite fit but, with work, could fit the new RFP. The big trend is capital markets analyst roles are being automated with AI. 

Embedded lending for Europe

2. Proov AI - An AI to validate AI model compliance

Proov allows financial institutions to demonstrate compliance with AI safety and fairness regulations. The LLM helps identify potential bias, disparate impact, and vulnerability in machine learning models. The platform saves significant analyst work and creates documentation for internal teams and auditors. 

🧠 This potentially democratizes the use of AI for smaller financial institutions. Not every FI has Chase's tech budget. If model validation can be even 50% more efficient, that's a huge unlock. You'd still need humans in the loop to review outputs. But the AI is a co-pilot. I wonder where these AI point solutions will eventually roll up. 

3. Agent Smyth - Stock Analyst AI

The platform analyzes real-time ticker and sector data to spot macro market trends or individual stock-level insights. Its agents provide traders, analysts, and portfolio managers with the insights to take action. The platform can provide pre-market prep in 30 seconds, ticker analysis in 2.5 seconds, and a red team for active trading strategies in 45 seconds.

🧠 A new AI-enabled generation of funds is being born. With all of these AI platforms, some AI is better than no analysts. I imagine the first-best customers here are emerging managers and funds. But every giant fund and asset manager started as an emerging manager.  

4. Lucra - KYC and Fraud detection for Law Firms and Funds (UK)

Lucra aims to eliminate the manual verification work that law firms and funds perform when performing KYC. It automatically pulls data from open banking and credit agencies to summarize a potential client's fraud or AML risk.

🧠 Funds and law firms rarely have a digital onboarding portal, but KYC is everywhere. KYC is becoming something all businesses have to do; even your accountant (CPA equivalent) has to do it. They're unlikely to deploy a website or mobile app to capture the needed data, but they do need to perform the checks. Automating that makes sense but points to a broader theme. KYC and KYB opportunities are as big, if not bigger, outside Fintech than inside.

Things to know 👀

1. Nubank hits 100m users

That makes them the first non Asian digital bank to hit the milestone.

🧠 I’ve written about Nubank before (including in the recent Brazil piece), but what stands out about this milestone, is 80% of those customers are monthly active. 

🧠 For revenue growth they have continue to cross-sell to the existing base but user growth will tail off without market entry or expansion.

🧠 With 92m of those in Brazil, the jury is still out if they can do market expansion. They’ve crushed. But I’m not sure they can continue their user growth rate at the historic average.

🧠 If they do manage market entry, we’re witnessing the creation of new global brands. Mexico is a much harder market, with much more competition, Colombia is waking up, and Nu could become a regional mega brand.

🧠 Will they enter the US? Given the rise of cross-border and inter-Americas trade it seems rational. But it’s so very very hard to become or buy a bank in the US, so would they go with a sponsor?

🧠 If I were them, I’d avoid Asia. APAC is another universe, filled with wallets, QR codes, countless payment types and well funded incumbents. While the dynamics in a Vietnam or Indonesia seem similar (underserved working class), the market reality is very different.

🧠 If you’re a bank CEO and you’re not obsessed with the Nubank case study, I have to ask; why? There are unique dynamics at play unquestionably. But understanding their execution and why it’s different to yours is critical.

2. FIS launches a BaaS platform

Core banking and payments provider FIS announced their “banking as a service” offering for banks called Atelio. Pitched as a way for banks and developers to have building blocks for launch innovative new financial services, the launch follows the acquisition of Bond in June of last year. KeyBank, College Ave and RoyalPay are “lighthouse clients” and FIS cited S&P data that “median sequential growth rate of (of deposits) 2.2% for banks, versus a decline of 0.8% for banks that did not”

🧠 The embedded finance business case remains a no-brainer. Smaller banks absolutely need a way to win deposits, and as US Bancorp results show, embedded finance is an effective way to achieve that with high RoE.

🧠 FIS has a distribution advantage. FIS has an estimated 9% market share of smaller banks (Kansas Fed) and a 16% share of all banks in the US (S&P). Smaller banks rely almost entirely on their core provider for technology, and have the greatest need for attracting deposits.

🧠 A distribution advantage is no guarantee of success. Bond was acquired for an undisclosed sum which suggests it did not have the most revenue at acquisition. Large companies like FIS have numerous products on the shelf, and various sales teams are trying to get them into the core provider.

🧠 Embedded Finance is becoming mainstream. Nearly ever large bank will have a running embedded finance team in the coming 18 months. This isn’t a trend going away despite the regulatory pressure. If anything, the opposite. It’s gaining momentum as more providers see an opportunity to solve problems.

Good Reads 📚

1. The CFPB Vision for a Data Economy

The terms and conditions for downloading an app doubled one man's car insurance. He had unknowingly agreed to "share data with 3rd parties." What happens when AI can see all of the data you didn't know you agreed to share? 

This brings us back to open finance and 1033. The broader idea is consumers should own, control and be informed of all usage of their data. Alex speculates that the vision of the CFPB is to expand the Fair Credit Reporting Act (FCRA) and 1033 to ensure all gaps are covered.

🧠 This is why informed consent matters. Your data is for sale, and it can have unintended consequences. AI can be biased or wrong, especially with poor or incomplete data. 

🧠 Regulations and laws don't always age well. The FCRA was passed in 1970, and since then, there's much more than employment and credit history available to CRAs (and everyone else). 

🧠 The EU had the right idea with GDPR; the execution was just staggeringly poor. Consumers should be informed of the use of their data and must consent to any usage. The implementation meant companies made dark patterns to create that "informed consent" tick in a box.

🧠 We should accept that data is useful and lucrative. If we reorient data around the device and the person and use tech like public key cryptography (PKI), we can do this. Data needs wallets, and it wants to flow between various AIs who can create value from it.

🧠 My bet? We'll see AI agents and agencies become data managers. The EU is even trying to build something it calls "data unions" to make this happen (which is the most socialist-sounding version of a good idea I could possibly imagine). We really need a place to store, move and manage data. Like a bank, but for data.

That's all, folks. 👋

Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)

(1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees. 

(2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a *. None of the above constitutes investment advice, and you should seek independent advice before making any investment decisions.

(3) Any companies mentioned are top of mind and used for illustrative purposes only.

(4) A team of researchers has not rigorously fact-checked this. Please don't take it as gospel—strong opinions weakly held 

(5) Citations may be missing, and I've done my best to cite, but I will always aim to update and correct the live version where possible. If I cited you and got the referencing wrong, please reach out