Kush on Kush's Blog

Moments Are Not Shared in Pixels

Kush — Wed, 23 Oct 2024 00:00:00 +0000

Read the following very carefully.

For just a moment I thought I was back on the second level, but I saw the room was smaller, there were fewer ticket windows and train gates, and the information booth in the centre was wood and old-looking. And the man in the booth wore a green eyeshade and long black sleeve protectors. The lights were dim and sort of flickering. Then I saw why; they were open-flame gaslights.

There were brass spittoons on the floor, and across the station a glint of light caught my eye; a man was pulling a gold watch from his vest pocket. He snapped open the cover, glanced at his watch and frowned. He wore a derby hat, a black four-button suit with tiny lapels, and he had a big, black, handlebar mustache. Then I looked around and saw that everyone in the station was dressed like eighteen-ninety-something; I never saw so many beards, sideburns and fancy mustaches in my life. A woman walked in through the train gate; she wore a dress with leg-of-mutton sleeves and skirts to the top of her high-buttoned shoes. Back of her, out on the tracks, I caught a glimpse of a locomotive, a very small Currier & Ives locomotive with a funnel-shaped stack. And then I knew.

Jack Finney, The Third Level

If you read it consciously without skimming through it, your brain was forced to recreate the scene the author was trying to describe. You had to make the mental effort to imagine the eighteen-ninety-something scene by visualizing the details given in the prose. Your brain pieced together bits and pieces of what it knows about how eighteen-ninety-something looked like from the various movies you have watched, or pictures and paintings you have seen in your lifetime.

Sometimes, reading such vivid descriptions of events and scenes might not seem as effortless as simply watching motion picture or a photograph. Of course, If I just had to convey the information from the scene to you, I could simply show you the following picture and call it a day.

Train Shed at Bangor, ca. 1890 (Bangor Public Library)

But you wouldn’t have to use your imagination as much. Sure, the mental picture you created might not be historically or factually accurate, as you might have accidentally imagined the wrong kind of hat, or the face of the man with the big black moustache might not be how he actually looked. But, that would be it. You will still sympathise with the authors experience.

Mind Creates the Information

Most people I know of, are delusional about photographs. What often happens is, when you look at one of your old photos, you don’t exactly see what’s captured in the frame, but you see what happened outside the frame when it was captured. You might see an old family group-photo from a beautiful mountain-scape, but you don’t necessarily pay much attention to how your hair looked, for example. What you actually remember is how funny the van sounded, or how everybody laughed at that boy who slipped on a rock, or how you almost had to vomit from motion sickness in the curvy roads.

All that was needed for your brain to trigger those nostalgic memories of your moments, was a photograph which represents a certain time of your life. But if you pay close attention, you might notice, you had to recreate a lot of scenes in your head. Again, the dress of the boy you just recalled, might not exactly be the same as he wore that day. But it doesn’t matter.

You know what else made you do this exact mental workout? Exactly. Verbal descriptions. It triggered the same process. You see where I am going…

The Best Way to Make Others Feel Your Experience

So, when I was young (er than I currently am), me and my friend were trying to create a tree house in a remote junkyard at the border of our colony. It was remote. The living-being most frequently noticeable was a mongoose, and a few ants on the trees. We collected some old wooden planks that were thrown away in the junkyard, and some jute strings we found nearby (by trodding through what seemed like a hiking adventure). We tied the planks and put them up on a tree branch like a bench to sit on. I had the higher one, being the skilled climber I was. This arrangement which we were particularly proud of, was called our tree-house (and another secret name which I am not at the liberty to disclose).

We also had a different project at the same location, of creating a large pit deep enough that two or three of us can cram in and be “hidden” from the world outside. Unfortunately, it simply turned out to be a tiny hole that could only fit a small box. So we did that, hid a box which we ourselves never found again. Some kids probably saw us digging from a distance.

Reading this might have been a fun experience for you. Now I can finally give you the only video I possess of this place,

That’s it. That’s the only media I have captured from this place. But I only need to look at it once to remember everything I have to know. Had I simply given you the video, it might have been boring. Even more so if it was just my face taking up half the frame while I tried to pose.

It might come to you as a surprise In this day and age that all you need to make people feel your experience, is to simply tell them. Moments are best shared verbally, with a bit of visual assistance from a photo or two to act as a trigger for the memories.

Contrary to popular belief, I actually think bulk photos of events with just dozens and dozens people’s faces in them trying to pose aka being fake in what they deem to be the latest trend, is actually counterproductive. It creates too much information, so much so that the brain doesn’t use it’s imagination anymore. Essentially, disabling the entire process of vividly remembering the experience, or making someone do so even if it wasn’t theirs. It is simply one of those hundreds of occassions when you wore a pretty dress. Not much if you think about it.

PS. This was written to support my stance on Instagram, specifically against posting of self-portraits in an argument with a friend. I argued that if the purpose of showing your face in new dresses repeatedly was to create/improve others’ impression of you, then it is pretty ineffective in it. For the same reason that if I already knew you personally, it would make no difference to me how pretty you look in your next photograph from the gazillions of times I have already seen the real you. And if I were a stranger, I will still not form a complete picture of your personality from the cringe-worthy pictures you take of yourself without knowing you. Hence posting your pictures is useless. He argued that it was for sharing your moments with friends on the platform, which required some context to break down.

Thinking From Scratch

Kush — Wed, 02 Oct 2024 00:00:00 +0000

Here is a simple problem. You are given an array of positive integers representing the heights of different buildings (in sequence) with unit width. Now assuming uniform rainfall falls over the entire 2 dimensional city, and water gets collected between every building fully until it can overflow, what will the total volume of the collected water be?

How a course teacher approached it was using some strategy previously discussed in a class. They had two loops figure out the right and left maximums for each building. Meaning, every building has another building some distance to its left whith the maximum height in it’s entire left, and same for the right side. When the heights maximum on the left and right side of every building were stored in two separate arrays, another loop can simply compare the two limits, and subtract the height of each building from the lower limit to find how much water is there exactly on top of that particular building. Add all the water above every building, and you get the total volume. However I had a very different simpler approach.

Thinking from Scratch

We can start thinking of our new approach in the form of puddles.

The total water stored is just the sum of all the water stored in the puddles (which may encompass one or more buildings).
To get the volume in a puddle, simply start a loop from the beginning of the array. Consider certain variables to be the state of our current scan. We need to have mutable variables “current candidate”, “puddle volume” and “total volume” declared at the beginning of our program.
Assume the current candidate to be the first element of the array, and start iterating over the array. If the next element of the array is smaller than our current candidate, then the water stored above it is simply the difference between the height of the current candidate and the height of the element where we are at. Add that to the puddle volume.
However, if the element we got to is either larger, or equal to the current candidate, then the puddle is completed. Add the water collected in the puddle to the total volume, reset the puddle volume to zero, and reassign the element as our new current candidate.
Continue doing this, eventually there will be a point when our current candidate will be the tallest building in the array, and the puddle water gets added for every subsequent building even though it is now inaccurate. But we should not worry about this, since after the tallest building becomes our current candidate, it is never going to encounter any element equal to or larger than it’s height, so all that surplus is never going to get counted in the total volume.
When the loop is finished, we would have calculated the volume of the water to the entire left side of the tallest building.
Now start iterating in reverse, and follow the same logic. This will add the water from the right side of the entire array to our total volume.
At the end of two complete loops, we will have collected the entire water from all the puddles.

The one small exception here would be if there are more than one tallest buildings (meaning there are more than one buildings with the same maximum height), In this case, the water between the first tallest building and the last tallest building would be counted twice. We can simply fix this by creating a third state variable that records the maximum height attained in our first loop. The second loop will end as soon as any maximum height element is found (element with the maximum height we recorded in the previous loop).

You can try to visualize the algorithm with the steps I gave above. Notice how we only need less than two full loops and a handful of variables stored in memory to compute this in contrast to the given solution which required two arrays of same length to be stored, along with 3 full loops.

Counting in Binary

Kush — Thu, 01 Aug 2024 00:00:00 +0000

Talking from experience, a lot of mediocre teachers don’t introduce binary numbers properly at all. They instead give you an algorithm to compute the binary representation of any given decimal number and hand it out to you without knowing why it is the way it is. Today, we are going to understand that process, but first we are going to learn to count.

Counting

So, assume you are a toddler who is learning to count. You are given 10 different symbols, namely

$$ 0, 1, 2, 3, 4, 5, 6, 7, 8\ and\ 9 $$

and are told that in this sequence each subsequent symbol represents the next number, formally known as the successive number. So, we begin counting. We start with this symbol: $0$, then we say, by adding “one”, we get this: $1$, then we get $2$ and so on until $9$. But then what? We ran out of all the symbols, so how do we show the successor of $9$?

Well, the decimal system invented the concept of place values. Since we ran out of digits to count, we just add another digit to the front of the previous one, and start counting again. Until now, that place didn’t exist, so we could say we were counting

$$ 01, 02, 03 \ldots 09 $$ and now the “tenth” number is represented by adding one to the left side place, like so: $10$. Subsequent numbers can be counted again, starting from $11$ till $19$. After that, we complete the digits once again, so we add another digit on the tenth place (signifying that another set of 10 numbers has been added) to form $20$. We continue adding successive digits to the tens place until we reach $99$. Now how do we write the the successor of $99$? Well, we create another place which counts how many tens of tens have been completed. We completed tens nine times until now, and the successor of 99 will be the tenth ten. We also start naming every tenth ten as a hundred. When we count nine hundreds, we say the tenth hundred represents a thousand, and so on, each time adding a place to the left of the digits. All these places show how many powers of ten we reached until now.

If you observed our counting process carefully, you would notice that an increase in the rightmost digit signifies the increase in the factor of zeroth power of ten (just 1), the second rightmost digit increased the factor of first power of ten (tens), the third rightmost digit increased the factor of second power of ten (hundreds) and so on. Just by putting it this way, I am trying to make the place value system which we learnt in primary school more intuitive.

Thus, we conclude by saying that any number, say $222011182$ is just

$$ 2 \times 10^8 + 2 \times 10^7 + 2 \times 10^6 + 0 \times 10^5 + 1 \times 10^4 + 1 \times 10^3 + 1 \times 10^2 + 8 \times 10^1 + 2 \times 10^0 $$

Counting with Two Digits

Well, we essentially repeat the same steps as above to count in binary. Just that this time, you don’t have ten different symbols. You only have the symbols $0$ and $1$.

We start again with the number $0$. It’s successor is $1$. Now we ran out of symbols, so we do the same thing as we did before .i.e, add a digit to the left. So the next number is $10$, followed by $11$. Now we ran out of digits again, so we add another digit to the left to get $100$, followed by $101, 110, 111$. We continue this process like so:

$$ \begin{matrix} 0\ 1\ 10\ 11\ 100\ 101\ 110\ 111\ 1000\ 1001\ 1010\ 1011\ 1100\ 1101\ 1110\ 1111 \end{matrix} $$

Take a moment to observe that this is essentially the same process as before, we didn’t invent some magical representation. It is just as natural as before, just that we don’t have a lot of digits to count with. You might notice that every place in a number now signifies a power of two. Which means, the number 1100 is

$$ 1 \times 2^3 + 1 \times 2^2 + 0 \times 2^1 + 0 \times 2^0 $$

You can evaluate this sum to get the decimal representation if you want, or you can keep it just like that. It doesn’t really matter, they mean the same thing, just with a different set of symbols (or digits).

In fact, apart from the standard algorithm to convert decimals into binary, you can use some basic intuition too. If you want to represent the number 37, think of the largest number less than 37 which is a power of two. Well, $2^5$ is $32$, so the sixth place from the right is going to be a 1. Next, we are left with $37-32=5$. Now, $2^2$ is going to be $4$, so the third place from the right is going to be a 1. Next, we are left with $5-4=1$, so the first place from the right is going to be a 1.

The binary representation will be $100101$. It is really that easy.

This of course, is only valid for natural counting numbers. We haven’t touched integers or floating point numbers yet. However, the goal of this post was to only make you feel that the natural numbers in binary, are just as natural as decimals. If you want to learn more about binary representations used in computing, I would suggest reading Chapter 2 of the book “Computer Systems: A Programmer’s Perspective” by Randal Bryant. It’s a wonderful book.

The Chronicles of Kush.in

Kush — Sat, 06 Jul 2024 00:00:00 +0000

The domain for this website was bought by my father some time after I was born (in 2006) and created a simple website shortly after I don’t exactly remember the tech stack for obvious reasons. Later when I began to explore web development, I took over the domain and hosted some of my stuff on the subdomains. In 10th grade, I thought using React to build a math solving website would be cool, so I did. That thing probably has archaic code right now, thanks cloudflare for keeping it running.

BytePost

Unlike most of my projects that never get finished, my blog is an exceptional project that got finished not once, but several times over. I started out with Jekyll, because some jekyll theme probably landed on my github feed, and I thought it would be really cool to copy that template for a blog whose contents I was not entirely prepared for. Turns out, I wasn’t quite good at HTML and CSS at the time Although who says I am now XD. But after what seemed like a tremendous effort of borrowing someone’s code to implement a dark theme, adding an option to switch was a real achievement. More effort was made into thinking of a name for this blog than what posts to write in it.

While writing this article, I realised wayback machine didn’t archive the old blog, and I don’t happen to have access to the old code as well, so I guess an old screenshot from a chat of mine should suffice.

This was bytepost, my old blog. Hey, don't make fun of me this was 2021 and I was 15

This was just the blog, the landing page for my website was different. By different, I mean very bad. It was a cesspool of all bad practices one could possibly implement in a website (never knew about lighthouse back then).

After a significant time of sticking with this cringe worthy page, I came to explore the YavaScript world beyond React. Fireship made me use sveltekit, and what better way of exploring a new framework than rewriting my blog. So that’s what I did.

The Website With SvelteKit

Here too, I started out with some template, but ended up making so many changes that I would barely call it a copy. I watched a couple of Hyperplexed’s videos to bloat the entire website with a ton of css and fancy animations, only to gradually remove them one by one until it was minimalistic enough for me again. This is what I ended up with. In terms of design choices, it was pretty decent. It was also fast enough, as sveltekit has SSG.

This website stuck around for quite a while, BytePost had been removed and the blog and landing page were consolidated into one website. It had an aesthetic look and decent performance. I tweaked it several times to get perfect lighthouse scores, and everything was really good and stable.

That was only until I was made aware of the harsh reality of the javascript world, I had to keep updating my code for every little breaking change sveltekit went through. A lot of the things didn’t work as expected, $ \LaTeX $ wasn’t supported. And it was just a terrible experience trying to maintain a codebase which I wanted to forget about. I spent more time on tweaking the javascript for a dead simple static site than actually writing any content.

New Goals In Mind

I came across Tufte CSS a while ago, and really wanted a blog with all those features. I wanted them so badly that for a moment I was considering to write raw HTML like Tufte CSS’s example page for every blog post. However, I had other plans too. I had been looking at a couple of hugo themes which had tufte’s features. After several months of mere planning and exploring the idea was finalized. The following were roughly my goals for making this blog:

Negligible, or no javascript dependencies.
All the features and aesthetics of Tufte CSS.
Support for $ \LaTeX $
Blazingly fast page speeds and minimalism.

I did find a suitable template which satisfied the first three requirements. However, as I began using them I had to customize a lot of minor things to get what I wanted. In the end, I got a lot more than what I had expected when starting out which is a different saga altogether. I hope to cover the details of this new blog on a different post.

Bleed Skholē With a Thousand Cuts

Kush — Tue, 18 Jun 2024 00:00:00 +0000

The etymology geeks would know that skholē is an ancient Greek word used for a place of philosophical discussion and leisure that eventually became today’s “school”. However, in that sense, skholḗ is already dead, but we are killing the very idea that skholḗ carried for academia with a thousand cuts.

Prologue

I recently graduated from school after 15 years of total school education in the Indian K-12 system. With all the positive aspects of education, I have a few very critical arguments against some of the methods and choices of the Indian education system Several things I say here might be true for other countries, but I cannot claim certitude.. While there might be some crucial problems in the Indian system concerning infrastructure and quality of instruction, I will not be covering those aspects because I completed all my 15 years of education in a school which might be one of the few top CBSE-affiliated schools in India. I have seldom faced any issues with regard to the quality of instruction or school infrastructure when I take a relativistic point-of-view with the rest of India.

My complaints with regards to education will be something that must concern the respective authorities from education boards and government bodies like CBSE, and not poor execution of regulations in individual schools i.e., they will be focused a lot on academics, examinations and syllabi. Thus, all matters here will be valid even when considering a perfectly ideal school that follows every CBSE (or whatever state board) norm and has highly qualified teachers. I am writing this critique as a future engineer who wants to talk about the problems of the schooling system that hinder the creation of better engineers. Forgive me for the lengthy text, for I just have a lot to say.Due to the lack of experience and motivation, I will also tend to exclude a lot of criticism from areas other than STEM and science academia.

Another vital thing that I would like to clarify in the beginning is that I am a well-wisher and an optimist for education. A lot of people tend to blame external systems and entities for their failures. Even though I scored 85.4% in my final 12th grade board exams, I am not writing this as a form of frustration towards my failures in my studies. I have well understood my flaws, so I do not blame any other person or institution for things that I have done erroneously.

The Death of Curiosity

My driving force for exploration and learning has always been curiosity. In fact, in our culture, by definition, a student is someone curious to know. I seemed to be a little bizarre to my peers for paying too much attention to some topics discussed in class, with a will to go to great lengths to either prove my point to the teacher, or to spend several hours trying to understand something that interests me, even though it is not a part of the syllabus. Although I don’t have any formal assessment to prove this, I would like to argue that out of a toolset of important stuff like memorization, curiosity, IQ (whatever that might be), accuracy, etc., I have only two strong superpowers, curiosity, and the ability to understand things better than everyone else. There is no such thing I can’t understand if I am curious about it. Even though I have a lot of friends with impeccable memory, and outstanding accuracy in problem-solving, I don’t think I have found anyone who is as curious or can understand novel stuff thrown to them as quickly as I can.

The crisis with the school curriculum is that it dissuades curiosity and the ability to understand a concept from its grassroots. One of the most outrageous things for me is when a pupil or a peer refuses to explore a subject or concept with me. The attitude that “I will not learn unless asked in an exam” is unintentionally promoted by the school curriculum by asking a limited set of questions from a limited set of concepts covered in textbooks. For example, when I was learning single variable calculus from MIT OCW, I was astounded by how much attention was paid to every detail and derivation of common differentiation formulas. One of the questions in the exam was

State the product rule for the derivative of a pair of differentiable functions f and g using your favourite notation. Then use the DEFINITION of the derivative to prove the product rule. Briefly justify your reasoning at each step.

I was stupefied for our exams never ask these kinds of questions. Often, the textbook lays down certain formulae as magic followed by exercises which use them to form several questions where the magical formula needs to be applied mindlessly.

To their defence, NCERT has also provided formal proofs and derivations for several magical formulae, but then again, if the tests don’t ask them, nobody will read them. However, this does not mean I want NCERT to publish derivations and CBSE to ask them as 5-mark questions. Because if that were the case, students would mug up every step of those proofs and write them down in exams mindlessly. The issue we are trying to fight here is the “mindlessness” part of things.

A brave thing to do is ask some open-ended questions in exams that cannot be mugged up from NCERT or any other source and require an inquisitive attitude to problem-solving. We want questions that appreciate the reasoning of students rather than the result. In the end, an engineer is more concerned with reasoning, while the result comes with the help of machines, SOPs and documentation. The government bodies use several new policies to their defence for this argument, but those policies mean nothing if there is no incentive for the students to get out of their comfort zone of mindlessness. Until each student is required to be good at reasoning and imagination to pass an exam, all those curriculum changes don’t influence anything. The only incentive students have in the current system is that of test scores. Either you can pivot the incentive to other methods of assessment, or make it such that the current incentive i.e., test scores, requires curiosity and imagination as an inherent necessity.

One solution that I would suggest is making the exams as unanticipated as possible. No fixed marking schemes, no fixed set of preparation material, and no fixed question format. As long as the exams are predictable, there will always be coaching institutions and schools that curb reasoning and curiosity to trade with test scores through memorization and mindless application of a direct pattern of questions. An exam should be such that there is no way to prepare for it. Assessment should not be the goal of learning but just a gauge of learning. The current system facilitates assessment as the final goal, and learning just as a means to do so. Instead, learning must be the goal, and unpredictable assessment as the gauge for learning, so there is no way to directly prepare for it.

Whilst teaching, there should be no consideration for either the student or the teacher on the weightage of that concept on the next test, or the question pattern of an exam. Your priority is being the best at what is taught, not figuring out how the tests will be conducted, while simultaneously being curious enough to investigate by yourself. This should be the only way to score well on tests, not through preparation for the test itself.

No student is free to ask interesting questions so long as they are incentivised to solve a particular pattern of pre-defined questions for a specific exam. Institutions simply beat around the bush by saying “We encourage students to ask questions.” or “We ask the students to perform certain activities.”, though the naked truth is that all these things are bluffs that are used as cloaks for an underlying rat race. As long as the rat race is the goal, none of the gimmicks will work. The entire assessment structure requires an overhaul such that it no longer remains the goal of learning. Any institution This is my retort to CBSE and the IITs. that uses these phrases to justify its stupid educational policies is either wickedly hypocritical or straight-up idiotic and stupid.

If I ask a science student why they are learning a certain concept on any random page of the textbook and they say “Because it carries 8 marks in the final exam” instead of “Because it is interesting” or “Because I was curious to know”, you have failed and you should be mortified because you butchered the student’s curiosity as of that moment. The solution I have proposed is to not know how many marks it carries, whether it carries any at all. They can’t even know if they will be asked a numerical, a derivation, or an open-ended question that is not mentioned on the page itself.You made them believe that those 8 marks were the causation of knowledge written on a page and not the other way around.

Too Much Abstraction

Another thing that I have always hated is when I asked a certain question and was told “You will learn this after so-and-so years”. The current curriculum relies too much on magic, as I have stated previously. There are too many instances in syllabi that don’t have enough pedagogical clarity as to why something is the way it is, and why it is relevant. I am not demanding a complete theoretical and practical justification of the stated facts, even basic intuition is enough to please a student’s curiosity For example, I had always been curious about what terms like sine, cosine etc. meant on the calculator as a kid. Whenever I asked this, the usual answer was that they would be taught in high school, but if someone had made the effort to draw a right-angled triangle and said that the ratio of these two sides is called the sine of this angle, that would be adequate for me..

Even after being taught, what is deemed “advanced” math for JEE, certain things were still laid out without proper background and justification. Sometimes I enjoy reading NCERT text that examines why something is true, and why it is important. Again, those things are not covered in lectures because although NCERT has them, they are never asked in any exam. Why talk about something that’s undoubtedly not going to be asked in an exam?

Grant Sanderson from 3Blue1Brown does an exceptional job of explaining the origins and the derivations of certain ideas. He has stated in one of his videos that the goal of these lectures is to make the learners walk on the path of the mathematicians who would have come up with these ideas, such that the facts that they are learning seem so apparent to them, that they would have come up with. It is because of this, that all his lectures seem so refreshing and educating. It feels like a magician revealing his tricks to the audience. This is also the reason a lot of the comments on his videos say “I wish they taught me this way in college or high school”. The general rule of thumb is that learning a new concept should feel like a revelation, or a conquest, not some odd prophecy that is handed down by gods without proper context.

But There’s a Catch

Conversely, this is somewhat the opposite of what engineers generally do. Engineers rely on black box abstraction where certain procedures are considered as a black box that takes an input and gives an output, and the internal mechanism is often ignored for the sake of solving a larger problem at hand. However, we are discussing about future engineers here. We have to supervise two factors whenever we consider teaching or learning new concepts: motivation, and hard work. The point I am going to make here is that they are not independent parameters, and motivation can control the work. Hard work is only hard when a person doesn’t have the right motivation Notice that I am not talking about the amount of motivation, but just the right motivation..

Now this might be a little subjective, but I believe that even saying “I want to pass cutoff scores for so-and-so institute” or “I want to get placed with so-and-so salary” isn’t the right kind of motivation. The basic criterion for motivation to be considered right is that when you have it, any amount of work you do becomes fun. By fun, I don’t mean in a masochistic sense, which is often the case for the motivations I just cited. With these motivations, people often get the feeling of pleasure in doing hard work by hallucinating the success they were aiming for, i.e., being admitted to a top institute or being rich after getting the salary they desired. They are not having fun in doing the hard work but thinking of having fun when the hard work is done, while the function of doing the work itself might be somewhat pain-inducing.

To talk about the right kind of motivation, I should give an example of myself when I was learning to code. I was not following some course or textbook for the same, I was literally on my own and mostly relied on Google searches and sparse pieces of really great advice from some internet personalities. This is akin to someone stranded on an island with a broken boat and an eternity to spend learning how to fix it without proper resources. I was trying to fix, or rather upgrade some open source discord bot without having any knowledge about it at all. I learned to program while fiddling with that code to get my desired features working. However, if you think about it, the process was supposed to be quite painful, after all, I had zero idea of what I was doing and had a codebase of over 3000 lines in front of me to fix. But the motivation I had was that if I figured out what this bit of code does, I might be able to add the XYZ feature to the bot The end goal was to make the bot one of the best bots available for public use, and gain user-base, but that was not what I had in mind when trying to code.. Often the reward of such a micro-goal was quite gratifying, as I learned inheritance (in OOP) along with implementing a certain new feature of the bot. The actual process of doing this was fun, not just the result.

I believe that you do not require big aspirations to have the right motivation. I am not refuting that your end goal shouldn’t be career-oriented. I have a good analogy for you to get an idea of what I mean by the right motivation. Imagine you love aeroplanes. You have been searching about various aviation companies, their aeroplanes and their features as a hobby. Maybe your end goal is as grand as building one yourself or flying one on your own, but that’s not certain at this point. One day, you get the chance to visit the cockpit to learn everything inside. Now, if you start taking this very seriously as if your life depends on knowing what every single button does and if you forget anything, you lose your chance of ever becoming a pilot, then the very same cockpit you were fascinated about will become a stressful nightmare. However, if you look at this opportunity flippantly, then you will be thrilled about getting to know the cockpit from the inside finally after so many years of merely dreaming about it. This applies to engineering as well. If as a child, you were fascinated with machines and always wondered how nature works and how we trick nature to work for us, when you get the opportunity to learn about it, it will become a thrilling experience. However, if you look at this as a do-or-die situation where you won’t be able to become a successful engineer if you fail to remember any formula or theorem, then this same fascination will become a stressful nightmare. Hence, if you are captivated about knowing something and when you finally get to know it, you will have the right kind of motivation. In fact, in this case, you will have enough motivation that any amount of effort you make won’t seem difficult at all. Hard work won’t seem to be as hard as you thought.

Why Does It Matter

All this talk on motivation in a section about abstraction. The reason I am discussing this is because the black box abstraction will destroy the motivation I just talked about. Imagine if the pilot or instructor who took you to the cockpit said “You don’t need to know why this is the way it is right now, just know that it does a certain action” to almost every question you ask them, then you will be disgruntled and dispirited to learn any further. This happened to me when I used to follow programming tutorials and they said “Ignore this bit of code, for now just think that it does XYZ” I used to hate that so much that often if it happened enough times, I would quit following that tutorial entirely. A good mentor is someone who always tries his best to provide good intuition in response to pupils’ questions. I remember the day when I saw the video “Fun to Imagine” with Richard Feynman and how elated I was to learn about the jiggling atoms’ explanation for so many things.

Fun to Imagine with Richard Feynman

The derivatives of the sine and cosine functions were simply handed down to us in school as something to remember for future application. However, both 3Blue1Brown and MIT OCW lectures by Prof. David Jerison provide a very good graphical intuition for why the derivatives are such.

Garbage Information Tactic for Assessments

I have been accused of hating chemistry as a part of a STEM subject. I partly agree that I love math and computer science, followed by physics and I dislike chemistry as it is taught right now. However, I would like to add a bit of nuance to this. I don’t hate chemistry as a subject, but I hate chemistry as it is taught right now. The way it is taught right now is as follows: Visit the Python Library Reference, you have 2 weeks to memorize all of its contents. After two weeks, the following questions are presented to you in an isolated exam hall with zero reference or support material whatsoever -

Write an expression that can be used to find out the population standard deviation of any list of numbers.
Differentiate between shutil.copystat() and shutil.copymode() functions based on their functionality.
The expression random.randint(a, b) will give a random integer N such that
1. $a < N < b$
2. $a \le N < b$
3. $a \le N \le b$
4. $a < N \le b$
5. None of the above

Fellow software developers reading this will understand the problem I am trying to get to. Even a software engineer with 15+ years of experience will not be able to answer these accurately if they were asked these out of the blue. Even after trying to memorize the entire documentation, many of them will struggle with such particulars. They are not failed engineers, the problem lies in the nature of the questions themselves. A real engineer doesn’t care to memorize such intricate details from any documentation, because they are never restricted from referring to the documentation to begin with. If I took the last JEE or NEET chemistry questions and gave them to a chemical engineer, there is a minuscule chance that they score anywhere near the required cutoffs for students, because as a chemical engineer, they don’t ever have to worry about the stuff that’s been asked. There would broadly be two categories of questions on this basis. I suppose that a chemical engineer must be able to explain the mesomeric effect in carbon compounds at any point in his career, but I don’t think they will be required to know that in the laboratory, a manganese (II) ion salt is oxidised by peroxodisulphate to permanganate as per the following reaction

$$ 2\mathrm{Mn}^{2+} + 5\mathrm{S}_2\mathrm{O}_8^{2-} + 8\mathrm{H}_2\mathrm{O} \rightarrow 2\mathrm{MnO}_4^- + 10\mathrm{SO}_4^{2-} + 16\mathrm{H}^+ $$

because most of the time when it is the case, they will either be performing this reaction 15 times a week and remember it by experience, or will be given an SOP as an intern, that has all the details laid out in front of them. The first category of questions is so fundamental to their field of science that any person pursuing higher education should be aware of its basics and how things are derived from it, but the second category of questions are things that can be found in a piece of manual or documentation and have absolutely no reason to be memorized. I do appreciate the first category of questions, but the second category of questions is not a good measure of one’s competence in the subject, but rather memorization skills. The second type of question is what I would like to call a garbage information tactic used by the examining authorities. This tactic seems to be effective for filtering out people who are not serious about an exam, by awarding only those who made the effort to memorize things in great detail from the textbook. People who can memorize this stuff easily will find these types of questions favourable for scoring well easily. The reason I hate chemistry is because this is what studying chemistry feels like. It feels like trying to remember everything written in a piece of manual.

I would like to add a bit of caution along with this argument. This is one of those famous arguments used by many students to hate STEM content. You will know how many times ignorant students say “When am I ever going to use this?” as an excuse for not learning important things. For example, I once heard someone say

When will I ever be required to know the electric field due to a thin uniformly charged infinite plane sheet?

To a normal person, yes, this might seem like garbage information, but upon further inspection, one will understand that the process used for deriving this electric field is very fundamental to physics. We use Guass’ law to figure out the electric field. The kind of integrals we solve to find this can be seen several times in physics to derive various other stuff, and even Guass’ law itself is a very fundamental piece of knowledge, we must know how it is used in real life. Also, an infinite sheet is a good approximation for a normal-sized sheet when you need to figure out the electric field a few nanometers away. Essentially, I am trying to say that this is NOT garbage information. So, one must be cautious when trying to argue whether something is garbage or not.

In connection to my aforesaid point regarding caution, there is one special way you can turn useful concepts into garbage information by putting a very short time constraint in exams. For example, you can ask a student to derive the electric field due to a thin uniformly charged infinite plane sheet, and use that expression to solve some numerical in 5 minutes, or you could simply ask the student a numerical and give them 1 minute to answer it. In the latter case, when there is a short time constraint, and derivations aren’t a requirement, the obvious thing to do is to memorize 15 different formulae for electric fields due to different kinds of shapes and conductors, and simply plug the values in the formula which you remember. Congratulations, you have converted a useful concept into a garbage information tactic and made the students mug up formulae to apply them mindlessly, AGAIN. This issue is very personal to me because this was one of the major reasons I scored less in various exams. I have way too many instances of running out of time for questions where I don’t remember the required formula and waste time on deriving it, only to realise it is not worth the time and I should move on to the next question which might be simpler.

This tactic gives a false sense of proficiency in a subject in which you might not be competent. It is very easy to score well in chemistry because of this reason. You don’t need deep knowledge of chemistry to know what reagent is used in Gattermann Koch reaction. Whatever you need to know is given in the textbook, mug it up, score well in exams, and then forget it. This is a very redundant process, if it was done in a manner in which you can safely forget it after the next test, I don’t see why it was on the test in the first place. I don’t protest against the information provided in the textbook. It is good to know the reagent if you are interested in remembering it, but then why is everybody required to remember it for the test, if no one is required to remember it after the test? If they provided us with every single named reaction, the periodic table, certain obscure formulae and scientific calculator, and still managed to make the exam challenging, then I would call it a real assessment of the subject. You must have all the tools at your disposal, just like an engineer would, and then the only thing required from your end, is thinking. I don’t think chemical engineers remember 200 named reactions like cavemen because software engineers sure do not.

Epilogue

I cannot reckon the number of times I have engaged in a mutual rant with a friend about the flaws of the system we are going through. After numerous years of learning to cope through the dissolution of aspirations and determination, aka maturing, it seems my friends should have finally conceded with undesired reality by accepting their fates within the system. The desire to create change finally seems like a childish longing, irrational enough to never conceive itself into reality at once. Thus, rendering this entire post as a useless effort turned to wasted time and energy. However, apart from the practical reasons to jot down these arguments, I still have clung to those childish longings to create any change, be it tiny enough to go neglected.

My scrutiny of schooling here revolves around the central idea of ideal STEM pedagogy which solves India’s technological problems and fixes the morale of aspiring scientists and engineers in academia. Whilst writing this, my outlook was that of an elated student excited about science and technology reviewing this system. Any arguments I might have missed can be deduced from entering this perspective. Although Indians find solace in hearing the shallow stories of poor rickshaw drivers becoming IITians and building successful careers, I am still seeking Nobel prize winners or industry pioneers. Sure, for the poor population of the country, achieving the best career opportunities from sheer competition is an exceptional reward at an individual level, but I might be from a minority of people who courageously challenge the fame that the so-called prestigious government institutions wield.

The Tech Startup Hype Is Hurting India

Kush — Thu, 12 Oct 2023 00:00:00 +0000

If you come across a headline like “A 16-year-old Indian girl launched an AI company in 2022, it is now valued at Rs 100 crore” today, wouldn’t you be surprised? How inspiring! India is growing, the kids here are extremely talented. Or is it 🤨?

This Delve AI Thing

Pranjali Awasthi started an AI startup called Delv AI that claims to “aid researchers in efficiently accessing specific information within the ever-expanding realm of online content.” Whenever I hear bold and fancy claims like this, I always investigate what their product really is. Overuse of such claims in the tech industry has made me a big skeptic.

Delv's interface You can upload a PDF and simply ask questions within it's context Copilot in Microsoft Edge It does the same thing for free So what does delv do? It basically reads a number of academic documents (PDF files), and uses the context to answer questions. Even though there are hundreds of websites popping up to do this exact same thing already and even the built-in bing pane in microsoft edge answers questions from any website or PDF that you have opened, for free, delv has raised $450,000 in funding and gives an approximate valuation of $12 million. Which, in India, is a LOT. I mean, seriously, with that money you could literally buy a large mansion in a metro city.

In fact, I just figured out that the screenshot which they are showing as an example of “AI-Powered Clustering Engine” is not even from their product, it is an open source project called Carrot2

Carrot2 is an open-source project

Alright, now moving on to what their “tech” startup really is according to me. This girl has used a number of open source LLMs, or paid API’s from big AI companies like OpenAI’s ChatGPT, open source products (Carrot2) and made a nice looking website to upload PDFs and a prompt UI to ask questions to the LLM (although I wouldn’t really call it particularly “nice”, it has design flaws). In fact, I could even pay a freelancer to make this crap in a week’s time and it would be better.

From A Long Trail of Scams

Money-laundering fraudulent startups is nothing new in India (and even elsehwere). I stumbled upon an absolutely horrendous scam called “Qpi Techonology” that made utter ludicrous claims about using “quantum simulations to make high energy density solid-state batteries”. In fact, if you happen to visit their website, which looks more suspicious than a microsoft tech support scam, every other sentence uses the words “quantum” and “AI” in it. They literally boast about everything ranging from SSBs, cloud infrastructure, quantum computers, semiconductors and what not.

The sad thing about this fraudulent startup nexus is that the naive media showcases them as real innovation to create an optimistic narrative to lure foreign investors.

The Not-so-scam Startups

Alright, I get it. These are just scams that will obviously be exposed at some point. But what about the real startups? Isn’t India producing an insane amount of unicorns every year? What about the e-commerce, fintech, edtech startups that are booming up?

According to me, those are not wrong, but simply overvalued. I am not an economics expert, but even if they are not overvalued from the money standpoint, they are definitely overvalued in terms of popularity among students and youth.

If we look at the startup’s that have risen to fame in India, you will find that the big fish are usually e-commerce, fintech or edtech startups which manage to gather an insane amount of funding and the small fish are the DIY (jugaad) solutions to trivial problems that get highlighted on social media and shows like shark tank. Most of the big fish are the ones which focus on milking the 1% lazy rich Indians to deliver a gulab jamun and their diabetes tablets at their home via their iOS app, while the remaining are coming up with solutions to make you buy fake internet coins with a shiny UI on the new bullshit AKA Web3 (did I mention there was supposed to be a Web5 too).

Let me make it clear that I’m not an ultra pessimistic conservative who wants the good ol’ days back. Being a developer myself, I have just seen the outrageous reality of the much-hyped Indian startup industry. The reality is that most of these so-called startups are just fancy dashboard UI’s and a simple backend which wouldnt take 2 months to build.

Oh, but Kush what about the idea and innovation! Startups are not about the mobile apps or websites but the underlying innovative idea!

Partially true. I agree that you should not be too excited about opening a wood cutting business that your grandpa once did. But look, even things like e-commerce have reached a saturation point now. Whatever can possibly be ordered online is already being sold online. YOU SIMPLY DO NOT NEED ANOTHER COMPANY TO DROP STUFF AT YOUR HOME.

I consider value-creation as something way more important than the cleverness and ingenuity of companies like CRED to fill their pockets (which they haven’t quite yet). I see businesses as a tap connecting a reservoir of resources and money to provide valuable products in a streamlined way. Pause for a while and think, how much exactly will a company contribute to the value of the overall economy of India if the only thing it does is earn revenue from credit card overdues. And yet, look the ridiculous amount of fame they have got.

Startups As “Get Rich Quick” Schemes

Are you in a tier 3 college? Don’t worry. Studies don’t matter nowadays. Education is obsolete. All you gotta do is think about something which you can make everyone buy. It doesn’t need to contribute much in their lives. First, create the supply. Create a simple backend with a NoSQL database running on a SaaS platform because you haven’t learned SQL and system administration yet, a simple frontend using React because you don’t know vanilla javascript yet. Then spend your time in creating artificial demand. Make people feel how much they need your service. In fact, you don’t even need to convince the actual customers, just boast about what your stupid website can do in front of filthy rich investors. You don’t need to create a unicorn anyways, a couple million would do. Once you get them, sell the startup and bam! You have earned enough to be jobless for a lifetime.

This is exactly what thousands Indian students are thinking right now. You can’t question them, because the startup industry is growing in India. They have this false sense of pride and achievement with creating a useless startup. The government, the parents, your peers, everyone is talking about startups. I’m not against startups, in fact, I encourage them, as long as they are done with the correct mindset.

The Correct Mindset

Most startups fail because they are solving non-existent problems. What is your motive behind creating a startup? Is it a problem that has been genuinely bothering you for a while? Or is it what I described above? Are you seeing your startup as a solution to a legitimate problem or as a solution to your skill issues in getting hired? If you had a legitimate problem, and your startup fails, you wouldn’t cry. You have other ways in mind to solve that problem. Success of the particular idea wasn’t necessary, the necessity was in that problem getting solved.

The second question you must re-evaluate is how much money and resources you seek in return for the value that your startup generates. Does your startup really help create enourmous value for the people? If yes, then you do deserve to be a millionaire.

So Why India?

All this time I have been trying to criticize the startup culture in India. But why only in India, you ask. These kinds of bubbles and scams happen even in the big economies like the U.S, then why am I targetting the Indian startup market specifically? This reality check is for those Indians who are overly optimistic about the future of India. I’m not saying you shouldn’t be optimistic, but you must also consider the following facts:

We are so focused on the service sector and “coding” skills that our manufacturing sector is lagging behind all the major economies.
We don’t have a single prominent indeginous smartphone manufacturing company yet. We don’t even manufacture the required parts to assemble a smartphones and laptops.
Our semiconductor industry is way behind competitors. As a last resort, our government has to lure foreign companies to manufacture in India by providing huge incentives, and even they are hesitant to come here due to lack of skilled engineers.
Talking about software, we don’t have large companies that specialize in building software, like Google or Microsoft. All we have is huge consultancy and service related companies.
Our basic transport infrastructure is one of the worse in the world.
We are lagging behind in the race for EVs and sustainable development products.
We have very low agricultural efficiency compared to other countries.
We have no significant AI companies (that produce AI models, not use them)
Our scientists managed to land a rover on the moon for $75 million, while our military is considering in spending $3 billion on US-made drones that cost around $100 million per unit.
Most of our military equipment is imported, and out of date. Indeginous equipment lacks the quality of the international standards.
Expenditure on R&D is very low, and quality of research is very poor.

If you heard the headlines which I mentioned in the beginning of this article, why don’t you ever hear headlines like OpenAI debuts DALL-E for generating images from text, Toyota And Idemitsu Team Up To Make 621-Mile Solid-State Batteries A Reality, IBM makes major leap in quantum computing error-detection, China completes first operation of world’s fastest train that travels at 600 km/hour, NVIDIA, ASML, TSMC and Synopsys Set Foundation for Next-Generation Chip Manufacturing ever coming out of India?

Sam Altman once remarked that it was “hopeless” for Indian companies to compete with OpenAI. Although he later clarified that this remark was not about targetting the potential of India, and was rather more about the capital required to train such models, I would like to ask you this question: Lets say even if he meant that, what were you able to do? Even if he would have actually smacked you right on your face that it is not possible for Indians to make anything close to what OpenAI has made, what response would you have? Because deep down, we all know it is somewhat true seeing the current status of research and development in India.

I am putting the effort of writing this post, not to make the youth pessimistic about India’s future. In fact, I believe it is excellent. We might be one of the biggest, if not the biggest economy in the world by the end of this century. But, dear students, please don’t get distracted by seeing some useless startups earn hundreds of crores. Please don’t leave your studies thinking that you will open another unprofitable startup to deliver gulab jamuns at home. If you think you can solve a genuine problem, sure go ahead. Small startups will play a pivotal role in determining India’s future, but don’t stop there. Please don’t forget the reality of India’s research, and infrastructure. Unless we start focusing on the advanced, useful, bleeding edge technologies like the U.S. and China, there is no hope for us to compete with them. Be proud, but know your limitations, and accept the challenge to break them.

About Teaching Computer Science

Kush — Sun, 25 Jun 2023 00:00:00 +0000

I recently watched Grant Sanderson’s (the man behind 3blue1brown) lecture on Math’s pedagogical curse at JNM 2023, and I think that it is a very interesting topic which I would like to loosely extend to programming and computer science in general. He talks about how rigor is a gift to mathematics but it simultaneously decreases the pedagogical clarity if given too much focus on. This, he says is the math’s “pedagogical curse”.

Now, I totally agree with this notion and I love math too, but for the sake of this article lets focus on computer science.

Learning Versus Information

What’s the most informative document for programmers? Documentation, of course. But we all know how boring and painful reading the documentation is. While it is important to provide the correct information, developers are enticed to writing more exact documentation for their code. But the seperation between documentation and learning resources must be kept. There is a tradeoff between learnable and informative resources. I am writing this short post just to tell the readers that any tutorials/explanations I provide on this blog will focus on pedagogical clarity more than information dense text. While trying to provide abstractions and analogies, I might also give inaccurate details of some concepts, so please consider them as what they actually are, abstractions. This means that I will focus more on providing analogies and abstractions. You are advised to read books and documentation when actually learning the topic in detail, or using it in your projects, my job is to get you curious enough to dig into the details.

You Are Not Creative Enough

Kush — Sun, 19 Mar 2023 00:00:00 +0000

So today I had an existential crisis due to a single thought. You probably think you have infinite creativity like me. Everyone thinks their thoughts are not bound, right? Well, I just hit the boundary today. You are not creative. Or at least creativity is not what you think it is.

Wow, that was intense. If you freaked out, don’t worry. Take a deep breath and relax. Let’s play God. You have to imagine a new world or a new reality which is entirely different from ours. There are just 2 rules:

This new reality has to be entirely imagined by your brain alone without using any external tool that generates entropy.
The constructs of this world should not be borrowed from ours i.e., the reality should be 100% unique i.e., don’t copy

Seems simple eh? Well let’s start with some basic imagination

Playing God

The creatures here look like blobs and have no shape. Is this valid?
- Nope. You know what a blob is because it exists in your world. Imagine a world which had perfect geometric shapes for everything. If you lived there, do you think you would know what a blob is? In short, you copied.
- Now imagine a completely fantastical form of objects which is neither perfectly geometric and neither curvy or neither a mix of the two. Can you?
The speed of light in that world is exactly 1 km/h.
- You copied again my friend. How do you know energy propagates through electromagnetic radiation in the first place. Heck, how will energy be defined in that world, is it even conserved? Oh! and what if there is nothing like energy!
- Think of a completely fantastical form of an entity which propagates in a completely fantastical fashion. Oh no! what does propagation even mean? Will it even exist?
This world exists on a 2-dimensional plane.
- What is a plane in that world? Is it i hat and j hat? But what if mathematics had to be different. Can you think of a completely fantastical form of mathematics and logic from scratch that doesn’t use the axioms that we do?

Ouch

So your brain hurts now. It’s as if . . . as if it can’t just think anymore. It hit the boundary of imagination. Are you creative enough? I don’t think so. There is always THE BOX at some point because you can’t think out of THE BOX of our UNIVERSE. Anyways, to be open to possibilities, email me if you ever come up with something out of THE BOX.

TL;DR: Creativity is building stuff on existing knowledge. Out of the box doesn’t exist, there is always some box in which you are thinking, even if you don’t consciously realise that.