Skip to content

2021

Belief Arbitrage

This is a 3 part essay.

Part 1: Talent

In my teenage years, I first understood that the kind of family you’re born into gives you access to certain kinds of wealth e.g. money, network, habits/knowledge and social capital to do things you want.

Most people in my peer group and even among the adults I could speak to, used these resources to make their own life materially more comfortable.

This is precisely what I wanted: material comforts, but unfortunately, I couldn’t die and pick a family to be born again in. So I kept looking for other means and ways in which I could make it work for me.

So I went hunting. I read a few biographies, centered across genres but mostly dead people viewed from specific lenses of entrepreneurship, learning, politics and the like. I also actively sought out and spoke to people in their 30s and 40s in careers similar to mine.

I realised that sheer raw talent and finding a way to apply those talents is also a unique form of arbitrage. It makes people blissful in what we’d now popularly call flow state. The happiest people I met were craftsmen and had high mastery. They also had a reasonable degree of autonomy over their own shape of the day.

I was also in my early 20s by this and realized that I wanted to optimize for more control of my day and mental energy, and was more than happy to trade off some x% wealth over this.

“Eureka!”, I finally thought to myself. I’d at last found a way, a strategy - if not a playbook which might work for me.

The plan was arguably incredibly simple and had only 2 parts:

  1. be good at a specific craft, which is in demand (Machine Learning for me)
    • be good
      • implies that you need to be at least in the top 2000 people in your age group for that craft
    • in demand - implies that there should be at least 10K people in that craft already across the globe and different skill levels and a mechanism for them to move out i.e. retirement, outdated, or similar
      • people who matter should know that you’re a skilled craftsman
      • my work is recommended by the likes of Andrew Ng, Paul Romer and a NLP book which has sold over 1K copies.

I think it’d be fair that I executed this strategy of sorts to the best of my then-ability.

In hindsight, I could’ve definitely done 5-10x more but we’ll leave regrets/lessons for a separate essay to retain the cheery narrative nature of this first person essay.

If there is something you want to take away from this essay, it is this: Talent is a valid form of advantage, but not everyone can monetize/utilize this well.

Part 2: Genius

To recap, talent is one’s ability to do something extremely well. I also argued that talent is the only form or advantage which I thought I could build to get material wealth in life.

To me, the idea that Genius can be translated to material wealth is an axiomatic belief. So we’ll not dwell on whether Genius is useful or not.

In the late 2000s, a young man would show up every week with the same idea for approval from the management of this rapidly growing Internet startup. He had previously quit his job at an prestigious consulting company and closed those doors behind him by going to a smaller startup in a completely different role.

For the idea which he had a lot of faith in, he was willing to bet this career and if not, at least his future prospects at that company. His bet? A new Internet explorer toolbar, which would ensure that users could find Google Search more easily and quickly. The man was Sundar Pichai.

To many, that is and was a move of genius. This was a genius of not just invention, but also business, technology and user psychology. There is more than one flavour of genius for sure.

Let’s begin by understanding what we mean by genius first.

Our cultural understanding of intellectual excellence, or more popularly, genius is extremely limited – and dare I say, harmful. We limit these to inventors, mathematical prodigies and their ilk.

Alexey Guzey talks about this cultural misunderstanding here: https://guzey.com/intelligence-killed-genius/

I know a few people who I believe to be geniuses. What happens when I tell them that I believe they’re a genius? They all tell me that there are people smarter than them and that they’re “only pretty good at one or two things”

bitch this is exactly what genius is.

Genius isn’t limited to the ability to solve puzzles, or do rapid arithmetic, or Sheldon’s eccentric quirks.

Genius is having the ability, no – the advantage, to see things which others can’t see.

To me, as someone of not-superhuman intelligence, the question is this then:

“How do I make the leap from Talent to Genius?”

The answer is in the Alexey Guzey essay itself: genius is having a vision.

In sports, the line between talent and genius is often blurred. Is Michael Jordan a really talented player, or a genius for exploiting the blind spots of referees to his advantage?

It’s being able to make bets and act on them with unparalleled insight. It’s executing. It’s doing something which escapes the imagination of most people in the know. It’s a taste for what could work, without being bound by tradition.

As Arthur Schopenhauer summarized: Talent hits a target no one else can hit; Genius hits a target no one else can see.

Part 3: Courage

We touched upon a question of interest to us:

“How do I make the leap from Talent to Genius?”

I don’t have a complete answer yet. So let me offer the part of the puzzle which I am very confident about: Courage.

It’s a key ingredient, a necessary catalyst for one to make the leap from one to another.

A natural question follows, what do we mean by courage? How do you recognize courage?

Of course, there is the Die Hard version of running into a gun nest. That is a subset of courage.

Courage also has some other flavors. Many of us recognize them almost instantly in the form of honor and duty.

The other smell for spotting courage is the acting on something despite fear and/or knowledge of harsh consequences. Many outsized entrepreneurial and military ventures are built with this courage.

There is also a lot of courage which stems from not caring about what people have to say. To neither be bound by shackles of history or tradition, but also respect Chesterton’s Fence.

It frees you to learn about reality from other people, but not be bound to the reality of today. An ability to envision a future state which others don’t imagine.

Pressfield’s Resistance is the most dangerous element to one's life and dreams since its sole mission is to sabotage aspirations. The first step to fighting Resistance is knowing that you can, and you can win a battle here and another there.

That is empowering you to action. That is a form of courage.

It frees you to make bets that’d make others shiver in fear of uncertainty. It frees you to act in ways congruent to your values and aspirations. To seek truth, beauty and craft above all else.

For a craftsman mindset, to do their best work, the ideal state is where they can act freely. For that specific form of high agency, courage is a key ingredient, a catalyst. Without courage, no talent can truly do their best work. The leap is impossible from Talent to Genius.

Courage in any flavor: pride, duty, honor, self-confidence is a foundational belief.

This is a belief that not only is something achievable - but there might be a means and methods to do so. It is a specific form of Arbitrage against Reality. An advantage which you, my dear talented friend, now have over others that don’t have the conviction, or the courage.

MLOps for Startups

Start your development by writing the overall impact and feature overview in the Press Release doc and README

If your time to ship is more than 2 weeks, write a functional spec

In case of bug fixes, add bug details or link to Asana/Github Issues

Always. Do trunk-based development. Don’t restrict a deployment trigger to specific people. As soon as you are done, go ahead, deploy and let others deploy.

SERVICE DETAILS

DOCS

  1. Please provide API documentation for your service (e.g. via API definition)
  2. Add auto-generated engineering docs in HTML/Markdown
  3. Who is the maintaining team/person at this moment for the service? All maintainers and committers should be listed Repo README should include instructions to set up this repo for development

Service Component — DATABASE

  1. List down all the database changes, if you added any columns, removed any columns, added or removed any tables.
  2. What kind of indexes do you have? If you added a new column does it require an index? If yes, why? If no, why not?
  3. Do you make changes to records? Do you do frequent deletes or updates?

Service Component — MONITORING

  1. List all the services you own and list down each server monitoring parameters
  2. Alerts for service uptime, service performance degradation e.g. latency, throughput
  3. Alerts for service machine disk/CPU/memory — what’s the threshold and how are they triggered

Please include today’s screenshots for each of them e.g. StackDriver. We need to make sure that you have proper monitoring in place.

Service Component — DEPLOYMENT

TODO Add a step by step documentation on how to write your first service and deploy it in the dev/stage/production environment for your org

CODE STYLING

PYTHON

  1. Add a pre-commit with black, sort and flake8 to your code. Follow the happy path convention. Add type hints

  2. Start writing code by writing APIs, then tests and then implement code. We use tests as API docs

  3. Add proper liveness and/or readiness check for your Kubernetes deployment:

Liveliness check is for Kubernetes to know whether your application is running.

Readiness check is for you to specify when traffic should be sent to the pod. For example, if your pod needs some operations to be done before it can take traffic (such as downloading a dataset), your readiness check should send back 200 only once it's completely ready for taking traffic.

Docs:

  1. Kubernetes Manual
  2. Blog on Best Practices

DATA BEST PRACTICES

VERSION CONTROL

Use dvc.org and version all your datasets to Cloud Storage Buckets and include your DVC file in your Github repo

EXPERIMENTS

  1. Aim for reproducible experiments by re-using actively maintained APIs

  2. There is no specific tooling to do reproducible experiments, but consider using something simple like Sacred or Hydra

Politics at Work

This quick entry is partially inspired by a similar entry from Manas Saloi. Ofc, other folks have offered their opinions on Basecamp and earlier Coinbase.

The woke mob is shitting on Jason and DHH, but I strongly agree with Basecamp’s new policies. IMHO a company is supposed to create an environment where people come and do their best work without having to align to one side of the political spectrum.

Want to be activists and offer your 2c on every political issue? Sure, do that in your free time. The company slack channel is not the place for it. I am surprised that it is even a controversial opinion.

Even more extreme position that I hold: Society (+most) Whatsapp groups should not have political discussions. I am there to know why the water pump did not run as per schedule. I don’t go there to learn about Modi’s covid performance report.

(Just my thoughts. Does not reflect my employer yada yada.)