You're Wasting Time Learning…
Timestamps
- - Intro
- - How to know what to learn to avoid burn out
- - Bleeding edge tax
- - The life cycle of technology
- - Specializations are going to show up
- - The importance of core fundamentals
- - What's next?
Every few years tech promises something life changing and you're expected to drop everything and learn it. Most of it fizzles out and you end up with what existed before, plus a few pieces of the new stuff blended in. That's just how adapting in tech works. A lot of what we hear about on social media is hype and marketing, and it takes time and a lot of people actually using the new tools to figure out what matters long term.
Learning fatigue
A lot of developers who seem uninterested in the current way of development are really just dealing with learning fatigue.
- "tech burnout" often means being tired of investing time into stuff that goes obsolete quickly
- learning is not easy or natural, new frameworks, languages and tools are painful to pick up
- grief around AI "ruining" software engineering is likely sunk cost
- you invested a ton of time into knowing things
- AI can now generate a lot of what used to take real effort
- so there's internal resistance to picking up the new ways of working
It's two sided though:
- risk of learning the cutting edge and having it go irrelevant
- risk of avoiding all the new stuff and actually falling behind
With the right approach you can spend about 10% of your time learning new stuff and 90% executing on what you already know, getting really good at your craft. That 10% is enough to stay up to date, as long as you're focusing on what truly matters. The hard part is figuring out what truly matters when everyone is talking about every new AI tool.
This isn't new
This is a huge problem right now with AI, but it's been a problem in tech forever. Web dev for example:
JavaScript → jQuery → TypeScript → Angular → React → Vue → Next.js → Redux → Redux Toolkit → React Query → ...
You never get a break. Now multiply that by 100 and you have the current environment with AI. Knowledge used to have a 2 year lifespan, now it's more like a 30 minute lifespan because that's when the next thing comes out.
The bleeding edge tax
General rule: the more you're on the bleeding edge (the newest of the new, past just cutting edge), the more pain there's going to be.
- upside: you get to some tools earlier than others, maybe a competitive advantage
- downside: you're investing time, energy and resources into things that may not end up relevant
Kind of like early startup investing. You can invest in a bunch of early startups but very few turn into something big.
Think of Angular 1 → Angular 2. Angular was a really good framework for single page apps, good state management, connected well with the backend. Then Angular 2 pretty much completely changed how you used it. The early adopters (I was on Angular 1) weren't rewarded for committing early, they were punished for it. Everything you invested in Angular 1 had to be translated to Angular 2, which was a big lift. That was about 10 years ago but the same idea applies today.
Hype cycles
Web3 is another one. Still a valid market, and time actually let us see what matters since a lot of the scams and companies not set up to succeed fizzled out. Software took pieces from it, but not every app is built on its own cryptocurrency or blockchain.
The equivalent today is "AI native" apps:
- no search, it's "AI search" or "smart search"
- no user written docs, it's an AI bot referencing different resources
- MCP or other AI integrations, sometimes just an MCP server wrapping an old school API
My guess is a lot of this is marketing, rebranding stuff as AI features. If the hype dies down there's a good chance some of it gets rebranded back to regular things. I don't know the future, but based on how things have worked before some of it will stay and a lot will go away.
I also always think about Yahoo. It was a big deal back in the day. Still around, but not what it used to be and never reached its potential. Some feature rich, super popular tool today could be the Yahoo of tomorrow. Good chance some of the tools you used this last year won't be used 1 to 5 years from now.
Same with CoffeeScript, I never really picked it up, but it was big for a while and now it's pretty much obsolete.
Translating that to AI: custom GPTs, plugins, custom agents, MCP, various SDKs, CLIs and tools. Some of it is going to matter.
How I decide what to learn
When I hear about something new I almost try to ignore it, unless I hear about it more than once or start seeing it more often. There has to be enough reason to invest the time before I spend the time, because I don't want to make bets on the future.
💡 Tip[!tip]
My job is a bit different. My success depends on picking up new tech quick, distilling it, and making content about it, so I spend a lot of time on the bleeding edge because that's what people want to hear about. If you're a software developer there's no reason to be on the nightly version of something. It's going to bite you in the butt more times than not.
I think back to my very first job. It was right when ASP.NET Core was being developed for its official release, and I started building on it because it was the new thing and I wanted cross platform. I put a lot of time into it and ultimately had to switch to an earlier version because it didn't have the libraries I needed. Early adoption hurt me rather than helped me. I still make that mistake, but I try to make it less damaging now: I'll try new tech, but I'm not trusting it as the infrastructure for a serious app.
- spending a little time on new tech gets you the majority of the benefit
- a lot of it is the same stuff repackaged and sold a different way
- I can pick up new tech pretty quick after a lot of practice, but it takes a lot to get me to abandon a workflow, tool or language completely because I understand the risk more
Specializations
Look back at Web3 again. A bunch of new roles and software got created, and a lot of it still exists, but you're not working with it regularly unless you're in one of those specialized roles.
Same thing is happening with AI, specialized AI roles are showing up. Core software engineering is changing and not everything gets pulled into the core.
- explore the different subsections of software engineering
- if one really interests you, shift your focus and get very good at it
- the general purpose software engineer is becoming less valuable
- specialization paired with really good general knowledge can get you very far
Anchor it to fundamentals
Every technology has fundamental concepts underneath. When learning something new like MCP, ask how it actually works and what the primitives are instead of copying some code until it runs. The MCP docs are a good place to see those primitives.
Example: with Web3 there was a lot of cryptography, which interested me. That has crossover with web dev, security of data and the protocols you run into as a web developer. If I only focused on the surface level stuff I wouldn't have picked up any of the fundamentals that actually add value to my skills over time.
You're building the skill of learning new technology, not just learning one specific technology. The goal is a versatile engineer who can pick up anything.
Things change all of the time. AI is just making the change happen faster and it's a bigger change than most people are comfortable with, especially if you've done things one way for years. It doesn't have to be a 100% change though, take the pieces that actually help you and work them into your current workflow.