Rental Culture

IP Owners have more and more promoted a rental culture in terms of IP. AI is just the latest manifestation of “you don’t really own this—we just let you rent it.”

Gather around children and listen to this old man spin you a tale. I know many of you will find this hard to believe but once upon a time, people actually owned copies of the media they enjoyed. People owned record albums. People owned VHS tapes. People owned CD’s etc. etc. Now the holders of the Intellectual Property (IP) in question would have told people that in a very real legal sense they didn’t own the IP. That is if you had a copy of, for example, Animal_House on VHS what you had from a legal perspective was a license to watch the movie off of the VHS tape at any time you cared to. You didn’t have the right to charge people to watch the movie; you didn’t have the right to churn out copies of the movie to give to all your friends; you didn’t have the right to project the movie on your garage door (well maybe you did have that last right—legally sort of murky I think.)

For the most part this system worked fine. Books were expensive and troublesome to copy so if you wanted to give a friend a book, you’d literally give them the book effectively transferring that license to them. VHS tapes were a little less troublesome to copy. CD’s were even less troublesome and because it was a digital (as opposed to analog) media you could make exact copies with very little effort. What to do? If people can easily give away copies of their licensed IP and still hang on to it well the IP holders are leaving profit on the table aren’t they? Mind you I’m not passing moral judgement on the arrangement; I’m simply trying to idenfity the logic involved.

The IP holders saw the future in Napster. They didn’t like the way that Napster facilitated piracy but it was hard to miss the fact that people liked being able to access their music and owning it didn’t mean that much to them. So legal streaming services sprung up. Spotify, Apple Music, Google Play, etc. etc. They allowed you to consume IP (music, movies, books etc.) without having to purchase a physical copy of it. You were effectively renting the IP from the IP holders. Now the fiction of owning media was even more exposed for the fiction it always was.

So now we come to Large Language Models (LLMs). Because many LLMs were trained on open source code they give us the appearance of having expertise and human intelligence in creating new software. This is an illusion; LLMs are no more intelligent than a parrot simply repeating what it’s heard a human being say. But the thing is LLMs give us a more concerning illusion. They give the illusion that one can now rent expertise the same way that one can rent IP. I want to build an application in JavaScript. With an LLM I don’t even need to learn JavaScript. I tell the LLM what it is I want and boom there’s my app. And so I can rent expertise. Or so I think. True expertise—the sort of Bloom’s Taxonomy higher order thinking that’s hardest to teach can’t be rented. At least not in the same way that IP can be rented. It’s this illusion of being able to rent expertise that strikes me as particularly dangerous.

I once contracted for Ford Motor Company. I was constantly bemoaning the state of the software we’d create for them; low-quality, hacky, etc. A very smart friend pointed out to me that Ford is a car company. As long as cars are rolling off the end of the assembly line they (Ford) are making money. If the software is good then terrific. If the software is poor but works, also terrific. Ford isn’t in the software business—they’re in the car business.

It’s interesting to me that so many companies are claiming that AI allows them to create software with, effectively, rented expertise. The idea being that like Ford, they’re not in the software business so software development expertise is not critical to their success or failure as a business.

The problem is that you really cannot rent expertise. Either you build your own expertise or you rely on someone who’s built expertise but you can’t rent it. Ford has built expertise in several tangential issues (e. g. supply chain management, industrial engineering, and marketing among others) to the manufacture of cars not because they wanted to but because they had to. Software sits in a funny place; it has to be bespoke to some extent or it provides no value to the business. And the bespoke part is the expertise of human beings that tell you what you need to build.

Now some will say that the expertise that’s being rented in terms of LLMs is the expertise of building the portions of the software that are commodities. CRUD screens and such like that. But the law of leaky abstractions is not going to magically go away because it annoys us. We cannot ignore the lower layers of software that CRUD screens and other such commodity constructs hide from plain view. The security concerns (among other concerns) are still present no matter whose expertise is used to create the software.

And if you try to rent expertise you’ll find that not only will there be dangerous holes in the software you create you’ll ultimately be in the same position as people who rent music, books, movies etc. That is, you’ll be forever paying a monthly fee for your rented expertise and you’ll never actually own it. I think a good argument could be made that expertise is the only thing a business really does need to own. A lot of businesses, justifiably so, think of software development as a commodity thing (in the Wardley Mapping sense of commodity). I disagree. And I think that the illusion of being able to rent expertise is leading us in the wrong direction. Don’t misunderstand—LLMs are certainly a powerful lever to those who know how to use them correctly. But thinking that somehow you can bypass the need for genuine expertise is an expensive folly.