The Rise of the Software Composer
The New Mozart is an eight-year-old composing web apps
🔌 This post was originally co-written by Jake Levirne, Akshay Bhushan and I. It was published on Tola Capital’s Learn Blog on 11/13/2024. Its so nice, I posted it twice.
Developing software is hard to approach. It requires abstract problem solving, an understanding of syntax and semantics, error handling and debugging, control logic, and a breadth of context.
About 70% of those who start today do so with Python. It’s taken over all educational settings for good reason: it has simple syntax, is readable and has wide application from the web to AI.
Still, only 27M worldwide attain the level of fluency required to develop software professionally.
Compare this to learning English. Approximately 500M people are fluent. Globally, English is taught at an average age of 6. In countries that prioritize it, programming starts at an average age of 11.
Large language models (“LLMs”) are both the AI era’s new interpreters and great equalizers.
Our new reality is elementary school students can easily compose web apps using English, without knowing Python.
Watch eight-year-old Fay write in English a full stack Harry Potter web app using Cursor and Claude.
She composes each step by expressing her intent and letting Claude 3.5 Sonnet generate the code to implement.
She navigates the IDE like a pro. She centers the Harry Potter font, adds an input box, and integrates a Cloudflare AI bot. She laughs with glee as she wills her creation into existence.
This new read-evaluate-print-loop (REPL) is extraordinarily powerful. It is far more so than if she were still trying to understand if her list comprehension worked as intended.
And it’s strikingly different from traditional development training.
This feedback loop removes the fear of failure. With sleight of hand, you sidestep the need to initially master data structures and jargon. But, you still learn to specify exactly what you want the computer to do.
Grizzled veterans will argue it isn’t software development. And that’s precisely the point. Hair splitting doesn’t matter. Ideas are turned into code that machines can run.
The Rise of the Software Composer
At the time of this writing only 0.34%1 of the world’s population is a software developer. In the 13 years since Marc Andreessen’s software is eating the world prediction, these new kingmakers2 have driven massive innovation.
The other 99.66% have constrained creativity. Today, they cannot easily bring their ideas to life. Imagine if the power to create wasn’t limited to just 1 in every 300 people. What if more had the means to turn their ideas into software reality?
Software development has always been about composition. Its essence is the act of breaking down a complex problem into smaller problems and composing simple solutions to form a complete solution.
But, we don’t label people this way: developer has been the term in vogue since the late 1960s.
In software, developers are seen as builders. They construct systems by writing code to create functionality. And, the term developer emphasizes a solution based on a set of requirements or specifications. Historically, the granularity of the work has been low level, with a focus on technical execution and efficiency to meet user needs.
In music, we call people composers. They create original works by selecting and blending notes, harmonies, and rhythms. This forms a complete, expressive piece. They work with abstract musical ideas and existing structures. They aim to be creative and artistic. They want to make something unique using the available tools and ideas.
Software developers have evolved and transitioned from writing binary, to C and Fortran to higher level languages like Python and Ruby in the early 1990s.
New compilers and interpreters were the big breakthroughs. They democratized programming with a new abstraction. This led to the growth of the software industry.
Modern software ecosystems flourished due to the rise of the GNU General Public License (GPL), the rise of the Linux operating system, Apache and Mozilla. Open source software encourages community driven development and rapid iteration. Git and GitHub brought social coding to the masses and the rest is history: 96% of codebases3 contain open source.
Each inflection point is due to standing on the shoulders of previous giants. This inflection point, however, is different.
Fays are now legion
To be a software composer is to select, arrange and blend pre-existing components, libraries and APIs into a complete and expressive solution. It’s a focus on creativity and artistry. A new, high-level abstraction makes it possible.
It’s extraordinarily approachable. For example, over 7,000 people have joined the Software Composer community. It was started by Riley Brown, a creator and AI educator with a wide following across YouTube, Instagram and TikTok.
The appeal is wanting and creating personalized software.
Your knowledge of English combined with a few additional LLM powered tools (i.e. V0, Cursor, Replit, Firebase and git) give you composition superpowers.
Software composers are building more than just toys. Make no mistake.
In the Software Composer community, there have been full-fledged AI SEO tools developed, a Down Detector and a Personal Image analyzer which is generating revenue.
Other impressive outcomes have been achieved with this same software composition workflow and tools across Twitter. This thread showcases a minimal AirBnB clone, an Exchange UI, a custom video editor and more.
Education of this new paradigm will take time to spread. Here are two early impressive examples:
Carloyn Zhang is teaching her design students at RISD the fundamentals (syllabus). She detailed the experience in this October 8th Twitter/X post.
McKay Wrigley built a very useful set of targeted courses at jointakeoff.com with over 5500 subscriptions
The most interesting signal – beginners don’t know what’s hard. Their ambitions are bigger, and they use AI to figure it out.
It’s still Day 0
The Rise of the Software Composer is a new chapter in technological innovation. LLMs let anyone bring their ideas to life. It’s no longer just for a few kingmakers.
This higher level of abstraction allows millions more to create and innovate. Like eight-year-old Fay, it’s a shift. It expands possibilities for all ages. It makes software accessible and personal. Mobile computing lowered the barriers to entry in media and entertainment profoundly in a single decade. It’s not hard to envisage the same disruption and democratization now happening in software.
What was once complex and intimidating is now as easy as composing music in a familiar language.
And it’s just begun.
27M / 8B people per industry reports (e.g. Evans Data). These 27M people include various levels of expertise from junior developers to seasoned professionals.
Stephen O’Grady’s 2013 “The New Kingmakers: How Developers Conquered the World” explores the developer-led landscape and its implications.
Per the 2024 Blackduck OSSRA report.