What If Time Has More Than One Dimension? Who Would Really Want to Know?

Sometime around 15 BI — Before Internet — I became fascinated by octonions. Like most people who become obsessed with a new mathematical structure, technology, or idea, I immediately started looking for problems it might solve — a hammer looking for nails. One was whether octonions offered an interesting way of thinking about time having more than one dimension. I wasn’t proposing a theory of physics; it was simply an idea that fascinated me. More than forty years later, multidimensional time is still far from mainstream physics, but it refuses to disappear. This week, New Scientist ran a feature on this subject. Reading it made me wonder about something more general than physics: what happens when an idea, if true, would require us to change too much of what we already know?

We like to think that ideas succeed or fail primarily on evidence. Eventually they should, but people don’t encounter ideas without baggage. We have careers, expertise, and years invested in understanding the world as it currently exists. Knowledge itself has dependencies: one result supports another, which supports another. Changing something near the bottom of that dependency graph can invalidate an extraordinary amount above it. History offers plenty of examples, such as Wegener on continental drift, Boltzmann on statistical mechanics, Marshall and Warren on peptic ulcers, McClintock on jumping genes, and Schmidt on Göbekli Tepe. In all these examples, the evidence and arguments were strong, but they were resisted because accepting them meant rebuilding a field from the ground up, and for years there was no convincing mechanism to do it.

I also think of a hierarchy of change. But there is an interesting inversion as you move through a hierarchy. The person maintaining the existing system often benefits from its continued existence. A kind of “Who Moved My Cheese?” syndrome? The person senior enough to own the replacement may benefit from destroying it. Both may be following the money. One is protecting the value the existing system created. In a way, resistance and evangelism are the same instinct: protect, or grow, the capital stored in one’s own skills. This helps explain why radical change sometimes produces the strange situation where the person with the original idea and the person at the top understand each other rather better than all the people between them. Their incentives happen to point toward change, even if for entirely different reasons.

Then there is risk, which is more powerful than money. An old system can be terrible and still feel safe because its failures are familiar. If I retain it and something breaks, we suffer another known problem. If I replace it with my clever new idea and something breaks, I made the decision. The organization’s risk and the individual’s risk are therefore not the same thing. Academia has its equivalent: being wrong with everybody else is relatively safe; being wrong alone can be fatal to a career, while being right alone isn’t necessarily much better if it takes twenty years for everybody else to notice. Before asking whether an organization resists change, it is worth asking a more precise question: who is being asked to take the risk, and who gets the reward if it works?

My own industry, computing, has an extraordinary capacity for pretending to change. We invent new frameworks and libraries constantly, yet until now surprisingly little has fundamentally changed in thirty years. Much of the Web is still an application that parses, generates, and renders HTML: markup designed for simple documents, stretched into an application runtime. Meanwhile, we rarely reconsider the fundamental interface between human and machine. The early Web had richer competing visions — Ted Nelson’s Xanadu, scientific document renderers, spatial and hypermedia interfaces. They didn’t all disappear because they were technically inferior. HTML accumulated an ecosystem.

Then we have what I call the algal bloom effect: successful ecosystems can eventually suppress alternatives. Initially, growth signals success, but eventually the successful organism changes the environment and consumes the oxygen competitors need. Just think of Hadoop.

And now we have the brave new world of LLMs. And yes, this is a real change, and it adds a fascinating new form of resistance. A great innovation, and they have no career to defend, and no technology investment to protect. But LLMs generate based only on their existing knowledge that’s based on millions of examples that are weighted by the mundane, the common, and the conventional. Ask an LLM to design an application, and it naturally gravitates toward the architectures represented most heavily in its knowledge.

This doesn’t make LLMs hostile to innovation — as a tool in the right hands, they can be extraordinary. But an unavoidable and obvious asymmetry is worth remembering: the past has more training data than the future. In some sense, we have mechanized intellectual inertia.

So I return to time. It may have one dimension. Additional temporal dimensions may be useful mathematics rather than physical reality. Something much stranger may await us. That is not really my point. Imagine only that tomorrow compelling evidence appeared that our basic conception of time was wrong. The consequences would reach far beyond a single theory, meaning people would experience the same evidence very differently depending on what they had invested in the existing model. New ideas never compete with old ideas on an empty field; they compete with the old idea plus everything we have already built upon it. The question worth asking about a sufficiently disruptive idea is not only “Is it true?” It is also “If it were true, who would really want to know?”