Thread by @then_there_was

Last January I experimented with a book recommendation service.

I wanted to find a way to discover better books.

I found out that if you wanted to recommend nonfiction books, you needed to identify local problems.

What do I mean by local problems?

People seem to prefer nonfiction books that solve their problems (even if they're not aware of the problem).

You hear about this all the time. People accidentally stumble upon a book that changed their whole perspective.

Actually they just had a problem they couldn't solve (or didn't realize) and the book solved it for them.

Or it at least identified the problem.

Recommendation services like Spotify and Netflix work well, because humans general share the same interests over a long span of time.

However, while humans may eventually share the same problems, it's unlikely they'll share the same problems at the same time.

In other words, you might want to know how to program. And I might want to learn how to program.

But maybe you want to learn right now and I'd like to learn later.

The priorities don't neatly line up.

Humans also generally have varying levels of knowledge.

Maybe you know a little about programming and I know nothing about it.

It's difficult to test what a person already knows in a real environment.

Then I realized books are like instructions that have to be compiled by your brain.

When absorbed properly, a book will produce behavioral and perceptual changes in a human.

Books are like software in that respect.

I realized you could translate many nonfiction books as software that took into account an individual's level of knowledge.

Like Duolingo does.

But Duolingo is different from a book.

It compresses thousands of books into a single program.

And that program can do what a book cannot: It can gauge how much you need to learn and adjust its information accordingly.

It is an ever-evolving book.

What I'm trying to say is this:

I believe software will compress thousands of books for a narrow domain.

Like foreign language. Or exercise.

And such software will give you a pseudo-expertise—much like how everyone with a smartphone can travel without needing to read a map.

Large swaths of information that previously took years to learn will be compressed to minutes.

I believe humans will form a hybrid with software to help them navigate the complexity of the world with minimal resources.

However, I don't believe this will lead to equality.

Rather, software will amplify the minor differences between individuals and give disproportionate leverage to those best-equipped to take advantage of it.