Web3 Journey: Hierarchical vs. Exploration Learning, Core DeFi Skills - BG's Builder Bootcamp

Transcript

Introduction to Jeffrey (Rare Skills) and the Web3 Learning Journey

gone through how to build your first app. Actually, we went through the basics of Ethereum first, right, with Patty from Arbitum. We went through how to build your first app, the scaffold UI, uh what did we do? Overcolateralized lending, prediction markets, stable coins, ZK voting, how to build with AI. Once you've gone through that whole run, once you've done the speedrun, you know how to build apps, that your next stop is going to be with this man right here at Rare Skills. This is Jeffrey. rare skills. Uh I think you said it to me, it's all about math, right? And I think you'll get into it more, but uh when you're building with us, we're not the math nerds. We're like helping you get the tools and uh the initial education, but what you need to do is level up and talk to the math nerds

next. And that's where DeFi really shines. So uh I'll just hand it over to Jeffrey and you take it away. Thank you. All right. Thank you very much. Okay. So, uh, I'll talk about Rare Skills a little bit more in a second. I'm the founder, just so you have a name to attach to the face. My name is Jeffrey. Um, but let me get a sense here. How many of you, uh, have done one, you guys are all developers, right? Okay. How many of you have been doing web 3 development for less than a year?

Okay. How about less than two years? Okay. All right. That's good. So um what I'm what I'm going to be talking about is the learning journey for the rest of your time in this ecosystem because it's a very fastmoving industry. It sometimes can feel very overwhelming like there's a lot to learn and um I see some people you know spinning their wheels when they're trying to make sense out make sense out of everything. So I want to provide some guidance around that. But first of all what's rare skills? Um, so we're an education platform that's focused at on intermediate and advanced level developers. Uh, and something I'm particularly proud of is that we've trained people at the

What is Rare Skills? (Education for Intermediate/Advanced Developers)

very highest levels. Um, for those so canina is a competitive security platform and the number one on the platform was an alumni from us and we've even VPs of engineering at firms you've heard of have uh studied with us before. So um, what there's two ways that people engage with us. One is we publish content for free. Uh so if you just go to our website r wherekkills.io uh particularly our blog you'll see a lot of very detailed walkthroughs on uh various topics that are pretty fundamental to being a uh blockchain engineer. Now these are now these are free. So you just I'm not logged in or anything like that. So you just click read tutorial and you're off and away and you can start learning about what's relevant bit by bit.

Free Practice: GitHub Repos with Incomplete or Buggy Code (Exploits)

Uh the other thing that we offer if you just go to our homepage, uh I'll talk about this a little bit more later, but it's important to actually practice what you're learning. So we give a lot of um these are GitHub repositories with incomplete code or buggy code that you're supposed to exploit one or the other. Okay, so that's the first way that people engage with us. The second way is that for people who want to work with some of our instructors to work through the content, we teach the same thing that's in the content but on an instructor-led basis. So that's not free because we need to pay for the instructor time. Um only a minority of people do that, but I the content's going to be the same regardless. Okay.

Instructor-Led Courses (Same Content, Paid for Instructor Time)

So I'm uh not going to talk too deeply about our content. I just first of all want you to be aware that it's exists and that's pretty widely used, especially if you choose to go into um zero knowledge proofs. Let's see if I can pull that up here. This uh the Z the ZK book as we call it. This is probably the most used resource in the industry for getting into zero knowledge proofs. Uh but if you read our articles, we really put a lot of effort into making sure that they're as understandable as possible. So you here is a visual illustration of unis swap v3 and uh we're t we trying to say hey look as we change the uh concentration in liquidity how is that going to affect the how the price is manipulated. So um yeah, if you if you talk to people who know

Visualizing Complex Concepts (Uniswap V3 Liquidity Manipulation)

us, you'll know that we put a lot of effort into making things as understandable as possible because we value your time. Now um so that's where you find it. So the rest of this talk is about how to learn effectively. Obviously one way is to go to our website and learn, but you don't just learn by reading content, right? Uh so when people get into uh learning I think a very helpful framework for thinking about how does learning works is knowing that there's kind of two major ways that educators go about it. There is a hierarchical learning and exploration learning. And what what exactly is that? Uh well, first of all, before I educate you on the differences between

How to Learn Effectively: Hierarchical vs. Exploration Learning

hierarchical learning and exploration learning, I need to warn you that I'm biased because our school uses hierarchical learning. Uh but I still think you'll get something useful out of this. And I'll say some nice things about exploration learning later. But fundamentally, it's how do you go about learning? How do you actually know that you've acquired the skills that you care about? Well, you need to produce something, right? Because if you're spending all day just watching videos or reading uh content, you don't actually know if you learned anything. You want to have something to show for it at the end of the day. And there's uh two ways to go about that. The uh exploration learning camp says let's start building projects right away, right?

Exploration Learning: Start Building Projects Right Away (e.g., building a copy of Uniswap)

And that's what what you guys have been doing uh in this boot camp. Uh let's uh we want to understand how unis swap works. Okay, great. Let's build a copy of unis swap. We want to understand how um oracles work. Okay, let's build an oracle. Now um the hierarchical learning approach more talks about okay projects are cool but what are the fundamental skills that are generalizable across everything and if you really learn those first and when it comes time for you to tackle your projects you'll find them a lot easier at least that's the train of thought behind it. So when people are engaged in exploration learning it's hey I have a project that I'm trying to

Hierarchical Learning: Master Fundamental Skills (Generalizable Knowledge)

finish and by the time I finish it that's going to teach me everything that I need to know along the way. uh hierarchical learning. You can kind of see it in our uh homepage here, right? The it's just doing it's it it kind of can feel like I don't know some uh some forms of tutors I guess where you're you know just doing exercise after exercise after exercise. So you're working with an instructor who has a belief that oh you need to follow this particular learning path to get to your destination the fastest. Um and it will emphasize things like hey let's try to make sure we really understand ERC20 tokens. Well, do you ERC20 tokens? You can ask an AI to create it for you, right? So, and then there's the open Zeppelin library which you can just import.

So, do you really need to know how ERC20 tokens work? Well, the hierarchical school of thought would say, well, because it's so fundamental, you want to have that information be so much second nature that you know it like the back of your hand so that whenever you're reading code that uses ERC20 tokens or you are uh building something that uses ERC20 tokens, you're not spending any mental cycles thinking about how does an ERC20 token work? what functions are available, what's the correct pattern for transferring tokens. Okay, so the hierarchical learning would emphasize you need to spend a lot of time on ERC20 tokens simply because they're fundamental and you don't want to treat it as an afterthought. Okay, the um so projects in exploration learning is very much like, hey guys, we're going to learn by building this thing right

Importance of Fluency: Why You Must Know ERC20 from Memory

off of the bat, which you know, which is fine. Um, in the hierarchical learning, the emphasis is more on a project should be the sum of everything that you've learned previously. Okay, so we know some tokconomics. We know how ERC20 works. We know how vault works. I'll get to this in a second. Now, let's tie all that all together and then build a project. Okay, so that's where you will see things running a little bit differently. uh on I will say if you go to YouTube um or maybe like medium or something like that the exploration learning is you know what is what you're going to see most of the time because uh it's very interesting to see a project come to life right away right so that's what uh gets people engaged and often times gets people into the space so I'm definitely grateful for people

Hierarchical View: Project is the Sum of Everything Learned Previously

who create content like that um I'll just be sharing the school of thought that talks about um an alternate way of going about it okay so if I say knowledge gap does that you guys at least roughly know what I'm talking about there, right? You encounter something and you you don't really have an idea of what's going on. So, in exploration learning, well, this is somewhat easier to handle now that we have AI, right? Hey, I don't know what's going on here. to a particular domain. So we should study those first. Um let me kind of give you an example. So in our solidity boot camp, this is how we teach people to get to DeFi. So um the cross contract calls how ERC20 works in binary math foundations.

Hierarchical Learning Example: Path to DeFi (Core Fundamentals First)

We actually have people memorize all of the syntax here, right? We actually expect our students to be able to code an ERC20 from scratch. uh we expect them to be able to make a try catch call to another contract or a low-level call to another contract from memory. They shouldn't need an AI to do that for them. And the math stuff, there's some basic formulas that get used so frequently that we say, you know, you should just memorize them. And then when the student meets with the instructor, then the instructor actually tests, are they able to produce this knowledge from memory? And not only can they produce it from memory, can they do it fluently. So the idea behind that is that well I if this stuff comes naturally to you then this stuff will just feel like a very basic extension

Expected Fluency: Code ERC20, Try/Catch, and Low-Level Calls from Memory

of it. Everything you're learning just feels like just one step removed from something that you already know extremely well. So for example if you know how to bake cookies then baking a cake is not that hard for you. Right? So, if you're learning baking, it might be we would first focus on the fundamental skills, which is I need to know how to mix flour and eggs together and get the consistency right. Once I have that, then I'll probably bake the least challenging thing to bake, which is probably cookies, and then move on to something that's a little bit more finicky, like a cake. And that's the progression that uh we see here. So, it's not until somewhere around 12 weeks in do people actually start learning this. But the nice thing about it is by the time they get

The Payoff: Understanding a 30,000-Line Codebase in 20 Minutes

here, they can read the 30,000line codebase and understand it in about 20 minutes, which I I think that's pretty cool. Um, okay. So, I just talked about that. You know, obviously I'd like it if you guys take our boot camp, but if you're doing it on your own, this is what I suggest you do. Uh, if you you really want to commit this stuff to memory because it's going to appear everywhere. You put money into Uniswap, uh, you're going to be dealing with vault because it's going to give you LP tokens. Same thing if you put it into a and whatnot. Uh the inflation attack well you can just Google that or ask an AI but whenever you see a lot of vaults especially unis swap you'll see this strange code in there that seems to be doing something random and if you spend some time

Required Core Knowledge (Tokens, Vaults, Cross Contract Calls, Inflation Attack)

trying to reverse engineer that that's time that's not going towards understanding the protocol as a whole. So it's good to know in advance like oh yeah this is a very common thing I should really know that so when I see it I recogni I understand it in two seconds versus having to spend uh a few hours learning it all right and by memorize I'm actually serious about that because that's how if if you can recite it from memory then you've actually internalized it because things that are second nature to you are things that you've naturally memorized right what's a good example of this right you probably Maybe we're in a new country right now, but if you were in your home country, you would know how to walk back to your house if you were put probably within like a 2 km

Why Memorization Matters: Internalizing Knowledge

radius away from it, right? Because that's just something that you understand really well. Uh, okay, I need to open up VS Code and start a new project that you're not going to Google how to do that. You already know how to do that, right? I'm just going to click this, create a folder, and then scaff and then run my scaffolding program. Right? So, the obviously the danger here is that some depending on your schooling background, you you may you might have had memorization pounded into your head in in a bad way. Um, I won't I'll spare you the stereotypes, but you probably know what I'm getting at, right? So, yes, you shouldn't memorize that some stuff if you don't know what you're doing. Okay, I'm not saying just memorize things if if if it just feels like a random isolated fact to you.

But if you really understand something, you will have it memorized. So, the direct path to get to really deep understanding is, hey, these are the things I really want to memorize. All right. Um, okay. So, why do we use hierarchical learning? Well, first of all, in our program, we say, "Hey, we want you to be finished with this program by a certain time, and we want certain outcomes." In exploration learning, you're probably going to have more fun with exploration learning because you get to try things out. You see results fast, but you probably won't retain the things because you did a lot of things that you didn't understand what you were doing. Uh, we when we have a time limit, we need to be focused on outcomes. And usually people who are coming to us are willing to sacrifice a

little bit of fun to get faster results. So that's um you know like I talked about if if you study in a way where you're really trying to remember what you learn and what you're learning directly affects all the stuff that you're going to encounter in the future you it actually doesn't require that much time uh to learn in this way in case where does hierarchy come from right because there's a hierarchy to what you learn. I didn't explicitly explain that but hopefully it's clear. Okay. So, um but you know, exploratory learning is popular for a reason. You guys did a lot of exploratory learning here. Um for one thing, I would say if if if the way you explore a new subject is by drilling the basics in it and by memorizing these core facts, you're

Advantages of Exploration Learning (Avoiding Boredom and Engagement)

going to get bored very fast, right? You have to um you have to know, hey, is this actually a worthwhile thing to do or not? And I want to see, hey, what kind of projects am I going to be able to create? So, you know, it's totally fine to be building projects for fun. Just be aware that as you're building a project and you're doing a lot of stuff that you don't understand, you're not going to automatically learn it. Just being immersed in a certain place does not mean automatic learning. If you've lived in a country and not picked up the language for like 3 years, you actually see that's common, right? There's a group of hey, you group of expats, why haven't you picked up the language? You've lived here for a while. Well, just because just because you're surrounded by something doesn't mean you learn it

automatically. It still requires some deliberate engagement on your part, which is why we push for memorization because memor kind of hard to memorize something without deliberate engagement. Okay. Um, but you know, there's it's it's a totally valid way to do it. You just want to be aware that these strategies optimize for different things and that's what you're actually trying to optimize for. Okay. So, if you were to go off and if your next step, I'm going to sort of presume that if you're in this boot camp, you're probably interested and hey, you know, I I really want to get into this blockchain space and develop on Ethereum. Okay. So, here's how I would suggest going about it. just I've taught hundreds of people and seen what works. So here's my here's what I would do. This is what we do in our boot

camp and I'm just telling you this for free. Okay. Um something I say every single week to people who are learning from me is DeFi is finance and finance is math. Right? So uh at the core of it a a DeFi protocol is just a mathematical model of how someone thinks incentives should work. That's really it. And solidity is just an expression of that mathematical model. Uh oh, this text is a bit small down here. Let me make this bold so you can read it a little better. That did not help. Okay, maybe if I do this

Recommended Learning Path: Math, Core Solidity, Core Standards (ERC20, ERC721, Vaults)

little stuff that kind of felt useless in primary school, you know, like logarithms and exponential laws actually come in handy in DeFi. It's how we uh pick what what size numbers should be. But uh the good news is you don't really need that much math. But I would say, you know, if you've really forgotten your algebra or you have a non-standard background, then you know, you can learn from Khan Academy for free. Um, in terms of solidity, solidity is not that big a language. But the parts that I really recommend focusing on are crosscontract calls. So whenever you see a try call, this is something you really want to understand because what makes DeFi different from traditional finance is that it's

composable. Hey, I'm going to use your tokens with my router and that guy's lending pool. So you really want to understand the nuances of how contracts talk to each other because aside from that, Solidity kind of looks like C or JavaScript, right? So for the most part, it looks pretty familiar, but these parts are the weird parts. They're the important parts and they're stuff that you really want to make sure that you understand. And if you want to test yourself, just use this free resource over here. Okay. Um, by the way, this is in order. So math is something I would prioritize more. And then once you're comfortable with that like algebra you many of you are probably already comfortable with that but if not you know I definitely recommend brushing it up then core solidity with a particular emphasis on cross contract calls um contract inheritance and

function overrides are a good thing to practice just because it'll help you understand uh code bases that you're reading like if you look at a or compound there's like a million levels of inheritance in there and unless you understand what inheritance is doing it's a little bit hard to read it okay and then the core standards this is obviously kind of similar to that graph that I showed you earlier, but I'm just um putting it out here. Uh ERC20, I actually have people code ERC20s from scratch. It's nine functions and after a week of practice, people get it, but it's something that sticks with them for a long time. Uh this is also a helpful resource because not all ERC20 tokens are implemented the same way.

uh when I say a vault, so if you uh deposit liquidity into a protocol and it gives you an LP token back saying, "Hey, this represents your share of the pool," then uh that's a vault. Now, the most famous implementation of it is ERC 4626. I don't have people memorize ERC 4626 because it's a very big and verbose standard and a lot of people don't actually use it. Uh but the principle behind it is everywhere. So you if you understand how vaults work, then you're just going to automatically understand already 30% of any DeFi codebase that you read. Not any, but like 95% of them at least. Uh kind of unrelated, but this is common, too. So it's worth knowing how that works. And then once you've gotten all that, remember you want to know vaults, math, and and tokens before you start trying to

really master the protocols because the protocols are just the collection of these. uh in our boot camp by the time people get to this point they understand a protocol in about 45 minutes because they're just rearranging existing knowledge. Um but the way that I recommend people really dig into the code bases and really understand how they work is deriving the formulas that are inside of them. That's something that remember how I said you have to really deliberately engage with something to learn it. um for simple concepts like tokens and vaults you can memorize that but it would be kind of crazy if you could do unis swap even unis swap v2 from memory I mean more power to you if you do that but you I can't do that so there's definitely a few parts I would forget uh

Protocol Understanding: Deriving the Formulas in Codebases

but really what you want to memorize is what are the components that make unis swap operate the way that it does and if you have those memorized then you can just figure everything else out as you go now the unis swap v2 white paper is actually quite good it shows you all the formulas that are used there and You can look at a formula and say, "Hey, I know where this came from." Or you look at the unis swap v2 periphery code which uh calculates how big a trade needs to happen in order to hey, I want to put 100 tokens in. How many am I going to get back? That can be mathematically derived from xy equals k, right? It's x plus delta x * y plus delta y is equal to x * y. And then you account for the fee. And then from there, it's algebra.

It's worth going through that exercise and making sure that that's fresh in your head because when you look at that then you look at the solidity code you already understand the solidity code not necessarily because you read the code but because you already know what it's trying to do. So you're approaching code bases with a mental model of this is what it's supposed to do versus oh I feel like I'm getting knocked around a pinball machine trying to figure out what's going on here. All right. And once you've gotten past that point, you're definitely part of a small group of people. But um this is when you can start coding your own protocols. It's a good exercise to start from a empty repo and say, "Hey, I need to build a lending protocol or a staking protocol or perpetual protocol or something." And by the

Achieving Expertise and Becoming a Manager of Managers

time you're at this stage, you already have an idea of well, well, I know there has to be a vault. I know that there has to be a way for the users to make trades. I know that um we need a way to liquidate someone if they're getting uh into a risky position. Okay, so I'm just going to start coding those one at a time. And if you've already gone through the other four suggestions, this will actually not that be that difficult. and it will just show you that hey because you spent all that time mastering the more foundational things you're just use it's like Lego pieces just putting putting something together to make something more complex. Okay, so I've just gone over the my suggested study plan for how you uh kind of get to the point where you're more

inter late intermediate because if you're able to code a D5 protocol from uh without referring like especially these are more simple ones but I would say if you can make these uh without referring to a protocol you're in you're in the very late intermediate stage. So, this is a good goal to get to. And if you're focused and you're following the road map that I suggested, I think you can get there in 16 to 24 weeks depending on how much effort you're putting in. Okay. Um, but of course, it doesn't stop there. We 3 moves very fast and it's full of very passionate people and sometimes it feels very overwhelming. And we've only talked about DeFi here. Maybe you're like, "Okay,

Jeffrey, that's cool, but I'm more interested in bridges or ZK or something like that." And when people look at all of the stuff that there is to learn, oftentimes they get very overwhelmed. Like, okay, cool. I got the basic math down. Now I need to learn cryptography. That's even more advanced. All right. Um, I understood how to make DeFi, but now ZK feels very big and intimidating. Uh, one thing that I, this conversation comes up a lot, is when when faced with this overwhelm, I once every two weeks or maybe even once every week, uh, I hop on a call with someone who's asking for help. Jeffrey, what should I focus on so that I can like make money? What's the what what industry is going to be hot? Well, first of all, if I knew what industry was going to be hot, I would be

a VC, not an educator. And uh second of all, all of the hot industries tend to get arbitrageed away very fast. So back in like 2021, 2022, um I I could quote people this much money for like a hundred lines of code or something just because nobody knew how to do NFTs really. So I was like, well, that's a really really good deal. And this was not me. Some people charged even more. Uh but then of course people realize like hey this is actually not that complicated. Anyone can learn how to do this. So of course that nobody pays for that anymore. Um if you're familiar with smart contract auditing. So inside of the smart contract auditing side of the field there's these platforms where you can participate in and um if you find a bug you submit it you get

How to Get a Job (A Story about Building an NFT in Assembly)

paid money. Maybe about like two years ago, you could find very simple bugs and get paid a month's salary, sometimes even higher. Uh nowadays, those issues are so well known that you'll hardly get paid anything for them. So, um and even more so, back in 2022, I um what I would do is like, hey, you want to get a job? All right, let's build a NFT entirely in assembly and then let's implement the diamond proxy pattern which is kind of famously complex. Now it kind of sucks but they would do that and then an employer would look at them and like wow you must be really good at what you do. Let's hire you. So, it was a great hack that I could use with some people and

they would get these six-figure jobs after like six weeks of effort or something like that, which which is great, but obviously that trick doesn't work anymore, right? Because any shortcut to success just tends to disappear. So, I would not spend your time looking for shortcuts to success. Instead, what you want to be thinking about is, well, if I'm going to be in this industry, how should I live my life? It's really a lifestyle thing. uh don't pressure yourself like, "Oh, I need to master all of this stuff soon." All you really need to care about is what are you doing every day? Do you set aside some time to cover up the more fundamental areas that every area of web 3 use? Um and are you spending some time uh learning? Now, the nice thing about topics over

here is there's a lot of great learning resources, but when you get to more advanced things like advanced strategies for dealing with impermanent loss, uh then there's suddenly a lot less resource. And if you study it enough, then suddenly you become an expert on the subject. So I would say the way I suggest you live your life is spend some time mastering the fundamentals and spend some time on once you've become you know later intermediate then spend some time on frontier knowledge stuff that only a 50 people in the world know stuff that there's no tutorials for stuff that AI is giving you hallucinations on if you know that stuff and you know that well that's how you get an advantage if you're if you care about career and opportunities. Okay, I

was just kind of talking about this. So, this and you know, bless their heart, they read our ZK book and um like, man, I'm learning some advanced stuff. You know, I'm really excited about this. Well, yes, but but the content is available for anyone to read. You know, this is knowledge that not only is known somebody our institution has spent a lot of time trying to make it understandable to as many people as possible. What you really want to get to is I understand something that the AI don't understand. Not not as hard as it sounds.

And I understand something that very few people in the world understand. That's the strategy that I would think about if I was saying, hey, I want to work in the industry and I want to have an edge because um I'm going to just skip to this one because it um talks about what I was talking about here. Um some of you may be looking for to work in this industry, some of you may not. So let me just share my experiences as it is right now. Uh if you go to some job board and try to apply as a solidity engineer, it's pretty rough. Uh because if you haven't managed TVL, so like a contract that's actually managing money, it's it's very hard for to stand out in terms of your experience.

If you try to go the auditing route, well, even right now there's great auditors who having a hard time getting a job because it was an easy way to make money. So, a lot of people came in and now it's very competitive. So, don't just chase the things that are kind of like in your face like, you know, study solidity and get a six-f figure job, study um auditing, get paid a lot of money as an independent contractor. Uh instead, look for the stuff that doesn't have flashing lights above it. If you want to get your foot into the industry, like I said, niche topics are a great way. Um, some something like bridges and any like ZK offchain integrations. The um if you think

if you if you're working and you see opportunities like man, the tooling here sucks. That's if I mean Foundry is great, but a lot of tooling tends to be a little bit behind if especially if you're coming from JavaScript world. Well, try to build the tool, right? try to build a tool that solves it and you can oftentimes get a grant. It's actually it's not as hard as it sounds. Um, developer relations and solutions engineer roles come up a lot because uh, blockchain is a very social industry, you need to have points of contact and that's a good way to try to get your foot in the door. And if you are building, you know, the actual front end or the backend infrastructure or the DevOps infrastructure, web3 still needs that. And you can apply any skills

you have from the past to get those kind of jobs. Okay. Um, let me having I'm going to jump back to here and this is just random advice from now on and then I'll wrap it up on these last five points. Okay. Uh, like I said, I've worked one-on-one with hundreds of people. The biggest predictor of your success is if you finish what you start. All right? So, I've given you a bunch of cool strategies, but kind of none of that matters in comparison to this. Um, and I know it's not novel. There's there's there is literally no secret. And it's very hard nowadays because there's a lot of things that are clamoring for your attention. Uh, if you go to, you know,

when I say crypto Twitter, like, you know, our little bubble on Twitter where, uh, we talk about web 3, people are really trying to pull you in each direction to be, you know, get your attention. You're going to have to learn how and they're really going to try to make you feel FOMO for not working on a certain subject. So, you're going to have to resist that and uh, just stick to it. The way I say make it manageable is, you know, you don't have to commit to something for a year in advance. Just say, "Hey, I'm going to commit to something for one week and I'm not going to get distracted by anything else afterwards." Like, "Wow, that I actually hated that. I don't want to keep doing this." All right. Well, you only lost one week and you learned how to commit to something, so it's still going to be a net

Fighting FOMO: Resisting Distractions and Sticking to Your Plan

positive. And you probably learned something in the process. All right. Um, and along those lines, it's kind of related to what I just talked about. I know deadlines kind of feel like you're bringing school back, but they exist for a reason. They really help you avoid. In fact, there if you find, hey, I'm doing something that I don't really enjoy or I don't really see the point of this or it wasn't what I thought it was. Well, by the time your deadline hits, you have an escape route. So, now you're the sort of person who can finish what they start and not stick on something that's actually not working. Uh so self-management this is actually how I manage the company just going on a separate tangent is I say pick what you're you tell me what are you going to work on for the next x amount of time and by when are

you going to finish it because anyone who can do that in a loop is someone who can be a manager of other people and and if they can do that very effectively then they can be a manager of managers because they can tell other managers how to do it because the whole thing is like am I working on something that's worthwhile and and am I finishing it at in a sensible time frame that's really it there's there's there's nothing magical to So try to apply that to yourself. Okay. Um I'm also kind of hinting at here that you don't want to just spend your time reading content, right? You want to produce something, right? Uh that's that because if that's what people are going to pay you for and it's also a way to really force the information into your brain. Okay, now let's talk about content because it might sound like

Importance of Production: You Don't Get Paid to Read, You Get Paid to Produce

I'm knocking content, which would be odd because our company has produced so much of it. Content is obviously helpful. It sucks to learn when there's no documentation, right? Uh but you but you may come away from now here's the funny thing. If you like let's say read about AI and you read enough clever blogs, you can you can actually sound like you know what you're talking about at a party like yeah you know like um we first of all we had RNN's and then we had LLMs and then we have like the the bitter lesson you know about the bitter lesson from reinforcement learning you know you can kind of vibe like you know what you're talking about.

If you read a lot of content in web 3, you can get to the point where it really sounds like you know what you're talking about. So it may feel like you learned something and you did uh to a certain extent, but you didn't get to the point where you could produce something or do something of value. So that's why you really want that to be the goal for uh the learning that you do. You don't want to just sound smart at the end of it. Okay. Um yes, I talked about that already, so I'm going to jump over it, but like I said, It's a think of it as a lifestyle. If I say, "Hey, you need to get to the point where you understand unis swap v4 and you need to understand how optimistic rollup works. 7 months, boom, get started." Golly, that would be stressful. I'm not even sure that's

possible. Right? So, um, but if I say, "Hey, but you need to do it over the course of six years." You say like, "Come on. I I do that for 20 minutes a day. I'll get there in like two years, right? 20 minutes a day is nothing, right?" Uh so I mean it's it takes it's a good habit. I'm not saying it's an easy thing to do but we spend 20 minutes a day doing things that are not productive. So we can certainly say hey 20 minutes I'm just going to really lock in and study the subject. So when you look at it on these time horizons um big subjects become a lot less intimidating.

People overestimate what they can do. There's a saying people overestimate what they can do in one year and underestimate what they can do in 10. Now, in web 3, obviously our time travels a lot faster. So, I'm going to say people overestimate what they can do in one month and underestimate what they can do in one year. So, just keep that uh perspective. I mean, you know, I've only been around for 4 years, right? Four years is not considered senior, but this is a very fastmoving industry where people come and go all the time. So, if you can just stick stick to something, you're all you're you're automatically going to be in the top 10%. Uh networking. Yeah. Something else I tell people, hey Jeff, I want to get a job. Well, the biggest determining factor in that is who, you know, because I've seen a lot

of underqualified people get work just because they knew the right person. And I don't mean that in the sense like they were deceiving the other person about their level of skill. It's just, hey, you know, there's this founder guy. He's like, hey, I want to start a web 3 company. I need to hire a blockchain engineer. Well, what am I going to do? Start interviewing a bunch of people on a job board when I don't even know how to interview them? I'm just going to go with my buddy who I trust to not screw me over. So, I'm going to hire my buddy. But it turns out buddy doesn't have doesn't really understand the domain. He just knows some things about web 3. I see this a lot. So, now there's a saying that your network is your network. It's cliche, but it's true. Um,

put yourself out there. I've hired maybe half of the people who work for me through Twitter just because they were posting things like, "Man, this guy, he likes to tackle hard subjects. That's really cool. I'll send him a DM. And uh and and and I'm not the only person doing that because the last person I sent a DM to like that, he said, "Ah, sorry, Jeffrey. I just got contract work. If you had reached out to me like two weeks sooner, I would have been happy to do that." Okay. So, so put your stuff out there because a lot of other a lot of other people who are hiring are keeping an eye out for that sort of stuff. All right. Um Okay, let's wrap everything up, I think. Yeah, that's the last slide. So wrapping things up, I obviously belong to the hierarchical learning

camp. If you're new to a subject, I suggest drilling the basics by memorizing them and making sure that these are things that are generally applicable to the things that you're going to keep using. And then once you have those mastered, you can compose them together into more complex knowledge like D5 protocols. If you want to learn by building projects and say, "Hey, I'm going to just figure it out as I go," that's totally valid. And some people have employed that strategy with great success. Just make sure that hey this is the way I'm you know that this is the way I'm going about it. I'm going to stick to it and I'm going to use the exploration learning strategy effectively. Okay, you don't want to just try out a bunch of random different things. You just say I'm going to do the exploration learning way

Hierarchical vs. Exploration: Both are Valid Learning Strategies

or I'm going to do the hierarchical learning way and I'm going to maximize the benefits of that strategy and any deficiencies that it has is something that I'm just going to learn to cover for I do the exercises on the homepage. Obviously this is coming from a hierarchical guy, right? But I recommend that you work through puzzle. These are free, by the way. They're on GitHub. So, you literally just open up one of these things, uh, open the source file, and fill in the missing code so that the test passes. If you keep doing this, then you get very fast feedback about if you understood the subject or not. These things are designed to be finishable in like four minutes or less. So, you get a lot of nice positive feedback.

can also learn if you understood the subject or not because you can't pass the test unless you knew what you were doing. Okay. Um so that's the that's the exercises there. Practice what you read. Obviously don't just read it. Uh if it if the if the article is talking about something mathematical then translate the code to math. That's or yeah sorry translate the math to code if it's talking about something mathematical and then all this wonderful stuff I told you about learning philosophies and uh strategies for career doesn't matter in compared to consistency that's really the only thing that matters uh okay so if you have any questions you know feel free to reach me I I

actually make myself available uh this week I may might be a little slow in getting back obviously because this is one heck of week. I love it. Um, if you want to catch me, we have an event tomorrow where we're doing live smart contract hacking. So, you'll get to watch um some hackers. They're share their screen and they'll be trying to break into contract. So, all the stuff you've learned about security, you get to see in real time. So, please stop by there if you feel so inclined tomorrow night. And I think that wraps it up. Okay, I'll I'll be over there if you

have any questions or anything like that. Oh, yeah. Go ahead. Oh, wait. Can we get him a mic? Okay, first of all, I wanted to thank you. It was like an amazing talk. Thank you. Thank Austin, Patrick Collins, and every guy that is selflessly teaching about all this. Thank you.

And what I wanted to ask you if I had to frontier learning for instance account abstraction right now what path do you suggest like going through the latest Ethereum improvement proposals or how if you have a suggestion for this? Okay. Well, account abstraction is not that big a field. there's only a handful of proposals related to them. The point that I would try to get to is where you can code at least the simpler ones from memory because that means you really understood them. Now to get to that point, a lot of account abstraction protocols rely on, hey, how do we bundle up uh a transaction that

some EOA is trying to make and then send it internally. So you really need to master how do transactions work, especially when it's working from EOA to contract and from contract to other contracts. And then a lot of the count abstraction um things have validation that's not necessarily using the traditional signatures. So that's something in a hierarchical learning perspective you would want to say well what alternative validation schemes do these people lose uh do these uh protocols use and I'm going to really focus on learning those and learning meaning I can reproduce some information related to how these alternative um validate these alternative authentication mechanisms work. So they're made up of authentication mechanisms. They're made up of how contracts interact with each other. Uh those components are stuff that I would really make sure

that I understand in isolation and then uh bring it to together with account abstraction. Thank you. And one last question. I also watched like on on abdra ciphering about account substraction. There is thing like the entry point handle users up and I sort of grasp but if I want to take it deeper how can I research beyond what's available how okay the source of tr something I tell people the source of truth is always the source code you need to be able to read the source code for

these account abstraction protocols and if and understand why they are the way that they are right not just say okay I understand mechanically what it's doing step by step but why did they make this design choice versus another one and then you can read through the discussion forums but generally when you're at frontier knowledge you need to you can't be relying on AI or tutorials or something like that right you need to be able to read co like for example when we're studying virtual machines like we write content about virtual machines sometimes and there's no content on it what do we do read the source code right now you have to put some context and okay, why is it this way versus another? Well, we can't be sure, but we can certainly see obvious problems that

were avoided by doing it this way. So, that kind of being able to navigate that kind of uncertainty is the skill that you need to have frontier knowledge. Now, but to be able to do that though, you need you can't, like I said, you can't just be reading code and figuring out as it goes. You need to understand the atomic parts that it's made of like, oh, this is an authentication mechanism. This is a call data bundling mechanism. This was probably an optimization for L2s back when call data was more expensive. So that way when you're reading the code, you're reading it in chunks, not line by line. Thank you very much. Sure. Hello. Um, thank you for the talk was amazing.

Thank you. My question is I am not developer but in my case I am DevOps SR engineer. who can I introduce you know which will be the path for our mindset or our you know background in infra kubernetes and all the stuff is that's is that's a field you want to get even deeper into and become like an expert in it yeah okay I'm not an expert on DevOps I my only experiencing with that is managing DevOps engineers back in my web two days but so I'm just going to try to answer that question from a general perspective of becoming an expert in a field. So there's two

components to becoming an expert in a field. One is actually having the skills and and the other one is having your peers recognize you as such. So um if you want your peers to recognize you as such that means you need to produce something that proves that you have the skill whether it's good content or open source tools that you're creating some some kind of a verifiable thing like I'm solving up until now unsolved problems and because those problems went unsolved for a long time we know that they're not easy problems to solve and I solved it so that establishes my expertise now the nice thing is that that goes hand in hand if you actually solve those problems you know what you're doing and you also establish yourself as an expert So the way that I I've seen work with myself and with other people is if you're

uh one thing obviously you have to be an expert. But how do you be an expert? You need to focus on problems that as an industry people have a hard time solving, right? You you you don't want to be solving problems that already have a solution because that makes you a junior or intermediate engineer. Those kind of problems, you need to be able to solve them fluently, right? right? You should have no no problem solving traditional alerting mechanisms or uh you know ingesting logs into storage and things like that like the very well-known things you need to be good at. But then there's some uh existing things like at least to my knowledge with DevOps there's not so many templates that you can just follow because every company is different. So what sort of general ab solutions apply to that in the abstract? Maybe that's not a big problem and that could just be

How to Become an Expert: Focus on Problems the Industry Has a Hard Time Solving

me not knowing a lot about DevOps, but that's how I would be thinking about it. Thank you. [Applause] Hi, thank you for the talk. Uh so you mentioned about studying every day 20 minutes and I was wondering how realistic it is when it comes to math for example. I'm saying it from actually experience. So I'm currently trying to do that and every time I'm starting to I'm finding myself, oh [ __ ] I need to understand this. I need to dive deeper in order to actually uh practice the things that I learn. And from 20 minutes, it's going to an hour to an hour and a half. And was wondering if you have any tips for that. Yeah. Well, it sounds simple, but it's hard, which is why most people don't do it. So don't

beat yourself up if you find it difficult because everyone finds it difficult. If if you find it hard to commit to a course of study, um I mean, let's talk about the obvious things. There could be distractions. there could be motivation issues, yada yada yada. But let's let's take that out of the way. If it's still hard to stick to a course of study, then it's most likely because the course of study is uh above your level or the learning material is not structured well. With math, um I wouldn't you want to be more focused on actually solving problems, not worrying about, hey, did I understand this or not? You can apply the algorithms in math to solve things and then understand after the fact why it worked. uh if you for example people very bad example but people use the quadratic equation without knowing how it

was derived right so it's perfectly reasonable hey I want to become fluent in understanding how roots of polomials work before I understand them in a more abstract sense uh but you kind of have to balance that because some people some people are totally okay with just mechanically doing math and other people find that like hey I at least need to know what I'm doing so you need to find what works for you but 80 at least 80% of your time needs to be spent actually doing problems whether it's math or coding thing and the rest of it is just filling in what you need to know so that you know what to practice next. A small followup. So I feel like in math you have the theory and then you have the practical.

So when you say 20 minutes you're referring to practical 20 minutes meaning the theory is not in this time uh time that you're mentioning for the kind of math that we do. Well what what do you want to use the math for? A variety of stuff including those kind of stuff that it's for here. Yeah. Well, I would say um definitely finish like a college algebra or algebra 2 course cuz that's going to be used in every field in in uh in web 3. So just make sure that you've mastered that. Once once you're at that point then it's going to kind of depend on what you're going to do whether it's going to be cryptography. Then that would be more along the lines of number theory and group theory. If it's DeFi, you would probably want to know at least a little bit of calculus,

right? So, um, but honestly, a lot of what we do is algebra, actually. So, I would really make sure that you have that down and it just comes down to practicing algebra problems. So, you're using a program that's giving you the the problems, right? And you practice them and you become fluent in it. Oh, over here. Hi. Um, thank you for the speech.

It's very, it's very great. Um, it's good to realize like the different ways of learning. Um, my question is about this. I'm a person of like I like a hierarchy learning as well but it really takes takes time and I'm right now um a front end engineer in web two and I try I'm trying to transfer my um career to defy section and I found it's really hard to like to learn this um math and some financial knowledges especially when we have a limit of time and plus we have to learn this alone is really too painful. Yeah. Do you have any tips that to relieve this uh this

pain to to learn this unfamiliar things? Um yeah, maybe find someone to learn together on some community. That's one method I can I can I can think of. Mhm. Um well, I mean the answer I gave you is going to kind of depend on your situation. So just, you know, feel feel free to ping me in case I can't give you a thorough answer. But it's going to be painful, right? That's why so few people do it. But and the other thing is you don't need to finish your 20 minutes and feel like it's a success each time. Just having done your 20 minutes and saying I did it. That is the success. And then doing it again and again is the success. And thinking okay there's not enough time. I'm busy. But that's why you want to think in terms of 20 30 minutes over

Q&A: Making the Process Less Painful

the course of years. You there's almost nothing you can't do with that kind of a strategy. You already have a job. So you're in a good place, right? So, um, so that that's the that's how I would think about it. If you say, "Hey, I'm trying to get here in six months and I have limited time." I I don't know how you could succeed. I don't think it's possible. So, so it depends. Yeah. Thank you. Thank you. Okay, I think that's it. All right.

Thank you very much. [Applause] [Music]