Relics of Value
Dan Murphy, March 14 2024
OG Tik Tok
Misunderstood value
Probably the most popular blog post Jerry Neumann has written - certainly the most eye-catching - is Your Board of Directors is Probably Going to Fire You.
One of the core insights of that post is a fundamental misunderstanding founders may have about the value they provide.
The founder thinks “these people just gave me millions of dollars. They believe in me, they invested in me.”
And that’s true in many circumstances. Many investors invest not because they have so much conviction about the idea being pitched, but because they have so much conviction about the people pitching it.
The investors are confident that eventually the founders will figure out some version of the idea, or a completely different idea, that works.
But at a certain point this flips. Once an idea takes off and becomes a real, tangible, valuable thing, the founder is less important relative to the thing they created.
The evidence of this lays in the heap of founding CEOs ousted by their boards once the investors came to the conclusion that the person who created the thing was no longer the best person to lead it forward.
At bottom they invested in the company, not the founder.
The key mistake of the ousted founder’s thinking was to not recognize how their ability to impact the company had changed.
At the outset creating something new was the most valuable thing. But having already created a thing, the most important way for them to impact the company became to lead it forward.¹
And so slowly the primary lever for the founder to add value changed out from under them until one day the balance tipped from creation to management.
AI-mageddon
The most useful lessons generalize and this is one of them.
I’m a software engineer and it’s a very interesting time to work in that trade. In 2 years will it be a valuable skill to know how to instruct computers by writing code? In 5 years?
Some very smart people think not.
“Coding was never the source of value, and people shouldn’t get overly attached to it.” says John Carmack.
They point to AI powered tools like Devin and Marbilism that will likely at some point be able to do much of the grunt work that software engineers today do.
None of that is to say that software won’t continue to be valuable, or that the ability to create it won’t still be valuable.
But you mustn’t (can’t pass up a good opportunity to say “mustn’t”) lose sight of what the underlying source of value is and what the opportunities you have to impact that source of value are.
The underlying source of value is and always will be the ability to solve people’s problems or enhance their lives in some way. Software is one avenue for doing that, and historically one way individuals have contributed to software value is by instructing computers via code.
But that may be changing. Perhaps software will increasingly be created by AI tools with human language interfaces. And if that comes to pass software engineers may find themselves in a position like that of the successful founder, who must recognize the need to switch from direct creation to management.
The ability to manage the software generating tools, to identify where software can solve a problem or make life better, to make sure it is in fact valuable, usable, and correct - these skills are today associated more with the engineering manager, product manager or designer, but they may become the only ones that really matter.
Demand for what?
Many people point to some version of Jevon’s Paradox when they hear arguments like this. Jevon’s paradox deals with improved efficiency actually increasing demand for something rather than decreasing it.
A classic example goes like this: you have a crowded highway. You add four extra lanes to it and expect the highway to become less crowded. But it actually becomes more crowded than ever and you are very sad.
The highway eventually becomes more crowded because people realize the added lanes have, at least temporarily, made it less crowded. And so of course they want drive on the less crowded road. As more and more people do this, the highway resaturates despite the added lanes.
Applied to AI and software Jevon’s paradox would say something like “if AI does make it very easy to create software, the price will drop and demand for software will simply rise and as a result software engineers will be as in-demand as ever.”
Indeed some version of this has happened many times in the past. People went from laboriously writing assembly to using increasingly higher-level languages and increasingly productivity-enhancing tools and frameworks.
All of those developments have lowered the cost of producing software, but none of them has resulted in software engineers becoming less valuable or obsolete.
The increased efficiency made more and more software use cases economically viable to develop, and the relative scarcity of talent able to participate in that production kept wages high for software engineers.
So why would things be different this time around?
Phase change
First, what it means to be a “software engineer” may be a stark departure from today’s work at a text editor. It may consist more of making change via natural langauge prompts to an AI, QAing the result, and circling back with the stakeholders to make sure it’s doing what’s needed. In short, closer to what an engineering manager or product manager does today.
Perhaps “software engineer” isn’t even a job, it’s a hat someone without much specialized computer science knowledge puts on for a few hours here and there to get the software they need.
If a reduced cost of creating software does lead to increased demand it is the demand for software that’s increasing, not necessarily for software engineers.
Perhaps more software creation is self-serve, or as readily accessible as publishing a blog post is today. That is, you don’t need an “expert” to create it for you.
Another way to think about this is as a phase change.
You can lower the temperature of water quite a bit without there being much of a qualitative change in its character.
But beyond a certain threshold, 32 degree fahrenheit, there is a dramatic change: from liquid to solid.
Perhaps it’s the same with the efficiency of software creation. Various gradual improvements in efficiency over time didn’t really qualitatively change the nature of software engineering. You were simply producing code with increased speed and efficacy.
Could this next generation of software creation tools bring about a phase change in software creation, where it scarcely resembles its old form? Where a minimally technical person can rapidly create their own high quality software?
If so that would mean more demand than ever for software itself, and little demand for expensive laborers to create it via an antiquated, slow, and error prone process: coding.
Relics
Today I’m very excited about the future. I love creating software “by hand” (without much AI) as I do today. It’s an endless source of learning and creative expression.
But if that changes in the coming years I don’t think that’s any reason to despair, so long as we avoid misconceptions about what is valuable and how we can contribute to it.
Founders, engineers, everyone and anyone can get fixated on relics of value, what appears to be and once was valuable but isn’t any longer.
To do this we need to embrace change, not shy away from it. We need to be willing to adopt new and different responsibilities, learn new skills, and get back to fundamentals about how we create value: by solving problems and making life better.
If you are adept at solving problems with technology you will always have a place in society.
If you’re a software engineer you may just have to find a new way to do that.
Could your business benefit from expert software engineering, product development or team building? Get in touch at hello@rdnt.software or visit https://rdnt.software/ to get started.
Radiant Software builds exceptional digital products using Ruby on Rails, React, Node.js, Python or your technology of choice.
We can accelerate your product development process with rapid prototypes, A/B testing, and analysis of technical feasibility. Or leverage our experience to level up your engineering team with technical interviewing, mentorship, and training in modern software development practices.
Footnotes
- Of course this is simplifying things quite a bit. Even in a very young organization or product there is some need for management. And even in very mature businesses there is some need for creation of novel things.