Making peace with the ambiguity of progress

Is there an arrow of progress in our universe? Or do things change without any particular direction as a goal, like a dust particle engaged in a Brownian motion, bumping and tumbling along randomly?

I don’t think there’s an answer to those two questions. Our thinking is designed to box phenomena into neatly packed categories that capture only a slice of reality. In fact, that’s where the problem with philosophy starts. Even if we both use the same word – say “love”, “free will” or “democracy” – we usually mean slightly different things and these slight differences provide all the fodder for the philosophical debate...  Read the entire post →

Why are we rich but hopeless

1/ The world is more materially abundant than ever, we’ve eliminated several diseases, lifted millions out of poverty.

Yet, people aren’t reporting higher levels of meaning or happiness than before.

via Sustainable Degrowth Through More Amateur Economy

2/ Why is this happening? Rising income or material abundance does not automatically lead to a higher satisfaction. And not just at a global level, but also at a personal level. Why?

3/ What’s happening is nothing new. Humanity has always sold to itself the idea of progress. Any idea of progress, because it comes from our linear thinking, is always unidimensional but life is incomprehensibly multidimensional. ...  Read the entire post →

Life is fractal, but markets are square

I recently read Venkat’s synopsis of the book Seeing like a state, which I followed up by an excellent blog post titled The Meridian of Her Greatness. Venkat challenged people to summarize the most important ideas from that post in a tweetstorm. He said if it gets more than 100 likes, on Twitter he’ll give away $1. I thought it was a fair deal, so here’s my attempt to distill some of the ideas into a visual essay.

1/ When humans wield their power in the world, they are limited by the linear nature of their thinking. The best example of this linearization is the top-down planning of modern suburbs. Contrast this with how nations and states emerged in a bottom-up fashion. ...  Read the entire post →

What is truth?

A tweet-thread like micro-blog on a topic that I’ve been obsessing over lately.

1/ Whenever someone says “this is true”, or “I’m a truth-seeker”, ask them to first define truth. (Or if you’re asking this question, answer what evidence will constitute truth for you).

2/ Getting a hold of the definition being used for truth is especially important when talking about complex systems like business, politics, economics, ecology or essentially any field where you usually can’t just read error-free data from a well-isolated system.

3/ This privilege of substituting data with the truth is mostly available only to physicists. But even there, interpretations of truth can be widely debated – is 5-sigma a good enough threshold for declaring the Higgs boson to be true? Well, it’s anybody’s guess.

4/ The word ‘truth’ is bothersome because it’s ill-defined. If something is ‘true’, it won’t be debated. If something is debated but is ‘true’, how would you differentiate ‘truth’ from ‘falsity’? You’d use your subjective judgment to assess the evidence and then make that distinction. If you’d do that, so will everybody else and they can arrive at an opposite conclusion. (Much to your chagrin, they usually do). How can your truth be different from someone else’s truth?

5/ As you can see,

this ‘truth’ business is a slippery slope. I’d much rather prefer to use the word ‘satisfaction’ ...  Read the entire post →

The Reverse Turing Test and proof-of-human currency

Are you a bot? No, seriously how can you prove that you are not. How can you prove that you are not some sort of algorithm crawling YouTube videos trying to make sense of this world? And how can you prove that I am not a bot, that I am not one of those Google’s AI?

This question may seem funny, but I find it one of the most important one facing our generation. Actually, the question is less about whether you are a bot or not, it’s more about how can you prove that you are human.

If you prefer watching a video instead of reading, I’ve narrated the entire essay in the following 8 minute video.

This question’s importance is due to the fact that with increased automation, humanity faces the grave danger of going jobless. This has already started to happen in many industries: right from replacing drivers that make up a significant majority of any working population to doctors, lawyers, and artists. Automation is increasingly taking over all the jobs that traditionally humans used to do and the economic implications of this automation is that more and more wealth is getting concentrated in fewer and fewer hands. The rise of Google, Apple, FB can be directly attributed to tech’s recent dominance, which is now accelerating at a very fast pace, concentrating wealth in the hands of a few shareholders.

One solution for the jobless world is the universal basic income (UBI) ...  Read the entire post →

Why deep neural networks work so well?

Earlier, I had written about machine learning algorithms and how they struggle to do things that a 5-year-old can master: walking, speaking, and drawing.

This time, I go into much more detail and explore a particular type of machine learning algorithm: deep neural networks (DNNs). These brain-inspired algorithms are effective even on “natural world” tasks: translate between languages, drive cars and recognize cats and dogs.

Why do deep neural networks work so well? Where does their magic come from? I explore these questions + more in my new 16-minute video.

Hope you like the video. If you are active on Youtube, consider subscribing to my channel.

Logistic maps (and what they tell us about free will)

I’m told that people have started preferring videos over text, and personally, I’m a fan of Youtube creators like 3Blue1Brown. I’ve learned a lot from videos that take a difficult topic and explain it in simple words.

Inspired by this shifting trend, I’m exploring communicating my ideas and thoughts via a narrated video (instead of text). Here’s the first one on an equation called logistic map (and its relation to free will). I hope you enjoy the video.

This is my first video, so I appreciate any feedback you send my way (via Twitter or email, see below). Love it? Hate it? Anything I should consider changing in future videos?

Your company’s org chart is more important than you think

Startup founders have many biases. Some are classic cognitive biases that impact decision making, while others are specific biases that impact their product thinking.

There’s yet another founder bias whose impact is not felt for a long time. It occurs when founders assume employees think and act like them. The often repeated advice that “early startup employees wear multiple hats” is an implication of this bias. I remember I assumed that just because I was able to do multiple things (coding, design, marketing, etc.) I expected our sales folks to make their own presentations and engineers to think of new product features.

It was a bad idea.

Wearing multiple hats is dangerous

Founders are all about breadth. Early in a company’s history, when the team size is just 2 or 3 people, everyone has to do multiple things. However, as the company expands and new people come onboard, the hangover of everyone doing multiple things remains. The first marketer does everything: from AdWords to writing blogs to web analytics to developing brand guidelines. Similarly, the first salesperson is expected to find leads, qualify people’s interest, setup meetings, give demos, negotiate, write RFPs and then try converting interested prospects into customers.

New employees who’re asked to do multiple things settle into these broad roles and give some level of performance. However, this performance is mediocre and a source of frustration early on, when either you’re not growing or you’re growing, but there’s chaos and confusion all around.

This is a terrible way to grow as a company

A founder has to realize that her org chart determines the limits on her company’s growth. People give an amazing performance when they’re given one well-defined thing to do. In a company’s early days when the hiring budget is limited, I understand that the temptation to hire for generalists is ever present. But generalists don’t give you growth (they’re great at experimentation though). Real growth kickstarts when specialists are brought on to do killer execution on things that your company can benefit most from.

Org chart should implement your strategy

Right from the start, the CEO/founder should constantly be thinking about the organization design that’s required today and may be required one year after. Nobody else would do this. No employee will come and say fire me, hire a specialist instead. A CEO/founder only has few jobs to do, and one of them is company strategy and by implication, designing the company’s org chart.

I define organization design as:

What roles should be there in the company and how those roles should be related to each other.

From my experience, many entrepreneurs and CEOs (blindly) follow industry norms in hiring and so their organization chart takes a standard shape that’s indistinguishable from their competitors. That’s inefficient because each company has an essentially unique strategy and hence deserves a unique organization design that implements that strategy effectively.

In some cases, org design happens by accident because there’s no well-thought growth strategy (“we will do better than competitors” isn’t a strategy, but this is a topic for another post). A prerequisite for doing org design is clarity on strategy because if there’s no clarity, whatever org you have will automatically start determining what your strategy.

Common mistakes in organization design

To be a good organization designer, you have to be a good psychologist. You have to first learn what conditions bring out stellar performances in individuals and then design a structure where people can find themselves in such conditions. Effective org design is difficult because the temptation to underinvest and the fear of bloated org always exists.

(I’ve learned a lot on org design from this blog)

Common mistakes:

  • Underinvesting in specialist roles. I heard someone say that you don’t realize how much better a job can be done until you’ve seen someone do it 10x better. This means that for every role in your company, there are people who can do parts or entirety of it 10x better than existing people. You don’t need a good marketer, what you need is someone who’s killer at AdWords when it comes to your industry. You don’t need a frontend engineer, what you need is frontend performance engineer who can speed up your app 10x and hence considerably impact user satisfaction. If there’s a job worth doing well (from the perspective of your strategy), hire a specialist.
  • Having quality functions report into quantity functions. Functions such as QA and development should always be parallel in org chart and not report to one another. If you report quality oriented functions into quantity oriented ones, quality will suffer. If you report quantity into quality, speed will suffer.
  • Having long-term initiatives report into people accountable for short term. Doing this is the reason why big organizations usually cannot innovate when it comes to completely new initiatives. For people who’re tasked with short-term targets, long-term initiatives are a distraction because at the start they’re simply too small or too risky to get meaningful attention or resources. Since they’re measured on short-term targets, the big and the scaled up is where their interest goes. This lack of early nurturing causes long-term initiatives to fail early, creating a vicious cycle of stagnation. To solve this, long-term initiatives (such as strategy, R&D lab or brand building) need to be put into a separate place in the org chart (perhaps under a leader who reports directly to the CEO).
  • Not eliminating outdated roles and functions fast enough. The org chart should change as the strategy of the organization changes, which happens automatically as the company grows. Org chart implements the strategy, so not changing it frequently means your company will keep attempting to grow via the old ways (which may not work because market shifts constantly). So one of the jobs of the CEO and the board is to frequently assess if the org chart is aligned to strategy. This is why there are about a gazzilion books on change management because people don’t like their roles to be redefined or getting a new boss or, in the worst case, their job being replaced or made redundant.
  • Promoting high performers to be managers and leaders. Super-hard to avoid in reality, but when individual contributors who’re high performers are promoted, the organization gets damaged twice: one, the person who does the specialist job really well isn’t there to perform it, second, you have a manager who is probably a mediocre one (when you could have gotten an experienced manager). If you promote your best performers to managers, ultimately your org will be full of mediocre managers. Too often org charts revolve around the availability of people (and the fear of losing high performers). The right way, however, is to be clear of what roles exist in the org chart and what types of people will perform those roles best. Don’t fit roles into people, fit people into roles.
  •  ...  Read the entire post →

    On the inefficiency of machine learning algorithms

    Let’s imagine Artificial Intelligence, but in reverse. In such a world, humans are equivalent of machine learning algorithms (like deep learning) and some aliens (or our simulation overlords) feed labeled information from their world into us and ask us to “learn” the mapping between the inputs and the outputs. In all likelihood, their world will be incomprehensible to us (as it would have a very different nature than our world). Hence, whatever data is fed to us will seem random (as, in our world, we’d have never come across it before).

    Because the data will look random, it’ll essentially be white noise. In this light, the task given to us seems impossible: how do we extract meaningful patterns from white noise?

    Imagine trying to associate this with the label ‘1’ in an alien dataset.

    Having zero background knowledge about the alien world, you’d not assume anything and hence will require lots of data points tease out any correlations present in the dataset. For example, maybe, when aliens mapped their data to our 2D visual field, the real determiner of whether it corresponds to a specific label or not is a combination of the 5th, 99th, and 213th pixel. Or, maybe when you save this white noise into a .wav file, it corresponds to a sine wave (which is the intended label). Or maybe, just to mess with us, aliens have simply fed us truly garbled data.

    The difficulty in finding patterns in the alien dataset is that, because we know nothing about the alien world, we’d need to have an infinite number of such hypotheses on what constitutes an input -> output mapping. This mapping could literally be anything and a few examples of data won’t suffice. We’d need millions of examples and a lot of patience to tease out patterns.

    For machines, ours is an alien world

    Now imagine what it would like to be a deep neural network trying to label images of handwritten digits. The MNIST dataset (see below) is one of the most famous datasets in machine learning, and, apparently, modern algorithms have achieved a “near human accuracy” on predicting which image corresponds to which digit.

    Images in the MNIST database

    But getting to that “near human accuracy” took 60,000 examples (the size of MNIST database) fed into the algorithm. A human child can start recognizing digits with only a few examples. Hence, the question arises:

    Why is the machine algorithm so inefficient as compared to a human child?

    The key to understanding the reason behind this inefficiency is to appreciate that what looks like a “1” to us looks like white noise to an algorithm. Algorithms have absolutely no idea about the nature of our world.

    Just like an alien data set will likely look random and incomprehensible to us. To an algorithm, the following two images are exactly the same because it knows nothing about our world.

    These two looks the same to a machine learning algorithm (image via here)

    The secret to quick learning in humans: evolution

    Evolution explains why humans are able to learn generic patterns using only a few examples. We haven’t popped into existence ex nihilo. Our existence is a result of several billion years of evolution where

    organisms who “knew” more about the world survived better than the ones who had no clue ...  Read the entire post →

    Evolution explains everything

    I love evolution. It’s hard to not get awed by a process that took Earth, a big rock full of chemicals, and gradually chiseled it to create humans, creatures full of complex emotions and behaviors. Impossible as it may seem, the mind-bogglingly diverse human behavior can be explained via evolution.

    Let’s take our sense of boredom. We dislike doing nothing so much that sitting still during meditation requires active concentration. We have this anti-boredom drive because our ancestors who were action-oriented survived longer and had more babies, ultimately outnumbering our ancestors who were happy chilling and doing nothing.

    Or take our compliant nature. We like authority, we believe in things that good orators say, and we take part in superstitions because, evolutionarily speaking, being unpopular is much worse than being wrong. Our ancestors who believed in true things that made them unpopular got less sex than the ones who happily became part of whatever falsehood bonded the society together.

    Lastly, take our worrying or anxious nature. Gautam Buddha called dukkha (or dissatisfaction) a core part of our moment to moment experience. This insight that we’re generally unhappy or dissatisfied with whatever we have also makes sense in an evolutionary light. Our ancestors who worried constantly and overplanned for even rare contingencies survived better than the ones who were complacent and happy-go-lucky.

    The book The Elephant in the Brain dives deeper into the topic of human behavior from an evolutionary perspective – here are my notes from it.

    We’re adaptation executors, not fitness optimizers

    Some people have the misconception that evolution optimizes an organism’s fitness. In reality, evolution couldn’t care less about you or me (as it’s evident by the constant dukkha in our lives). Evolution is a blind process that over time increases the incidence of “greedy” organisms that survive longer and have more babies by whatever means necessary.

    During evolution, small and random changes accumulate over several generations of organisms. These result in organisms with various sets of traits and behaviors. Some organisms may end up having a propensity to worry, while may be inclined to do nothing.

    Ultimately what trait ends up spreading in the population is determined by who survives longer and has more babies ...  Read the entire post →