Cracking the Code
Meet the new power lines.
Last week’s newsletter got me thinking about the huge consequences that a small change in code can have.
In his 1999 book, Code: And Other Laws of Cyberspace (remember cyberspace?), Harvard Law professor Lawrence Lessig debunked the idea that the internet couldn’t be regulated — a shocking idea at the time, because many people before him argued that it could be. Of course it can! All it is is just lines of code! Whether that code is used to create a place of freedom or oppressive control, he wrote, remained to be seen. (Hm…) In short: Code is powerful.
Lessig — from whom I was fortunate to take a seminal class (in fact, you can still see the paper I contributed to!) — also used to say that the structures of code are what form our society. For example, when chatrooms used to have a hard limit on the number of people who could enter them, that single line of code made organizing impossible within that structure. How are you supposed to organize a movement if only 10 people can come together?
We're seeing an example of that right now. Just take a look at X: Last Friday, they changed their terms of service so that they can train their AI models on your posts — possibly even if you opt out. Also, a small change in their business logic (which could literally be a single line of code) to create more “transparency” means that a whole slew of people are about to experience social media in a very different way. (I wrote about this update to X’s blocking feature last week.)
As a person who used to manage the teams who ran Twitter’s business logic, these are relatively small changes. But it’s kind of remarkable how small of a change to the system is needed to impact such large societal change.
But I can be the little optimistic bird, as Justin Hendix of Tech Policy Press described me on the podcast we recorded about AI policy and democracy. Just as single lines of code can have a horrific impact, they can be immensely powerful in good ways, too.
Nabiha Syed, the head of the Mozilla Foundation (where I’m a board member), pointed me to single lines of code that can have a huge impact. Meet Code Carbon, software that integrates into your Python codebase and gives you estimates of how much carbon it takes to run your code. It also shows developers how to lessen their code’s environmental impact.
I’ve written about this before, but AI has huge potential environmental impacts on the amount of power, water, etc., used to run and cool its data centers. I want to live in a future in which this is not the case. A number of major tech companies are investing in fusion power plants, etc., to make sure that we live in a world of abundance. (Following China’s lead, Amazon, Microsoft and Google are moving forward with their nuclear plans, while Meta’s plan for a nuclear-powered data center was blocked a couple of weeks ago after rare bees were discovered on the site.) Needless to say, we don't yet live in that world, and it's more than a little irresponsible to just assume it's going to happen.
The former engineer in me has to wonder: If you could build something super-efficient, then it may scale even more in a world of abundance. It’s pretty amazing to think that, thanks to software like Code Carbon, single lines of code could be literally more efficient than not. (Getting into the weeds for a second with the engineers out there: Refactoring can do a lot!)
We’re all so focused on coming up with big solutions. Nuclear power! Massive government oversight! But it’s important for us to look at seemingly miniscule changes that can add up to exponential change — from one little optimistic bird to another.
Worth the Read
ChatGPT has entered the doctor’s office, and studies show that it’s already outperforming doctors when it comes to diagnosing illnesses from a case report — even those doctors who had access to a chatbot.
The nonprofit TechTonic Justice released a report showing the sometimes devastating impact of AI-based decisions on the more than 200M low-income Americans when it comes to health care, housing, employment and education.
Could the China-linked Salt Typhoon hack help us get better telecoms regulation? Well, yesterday senators called it “breathtaking,” and called on the federal government to galvanize action now.
Very proud of my alma mater: MIT just announced it will offer free tuition for families making less than $200,000 a year.
Also proud that Speakeasy AI, which I started and sold, is mentioned in Fast Company’s piece on the 5 best North American tech innovations of 2024!