Why did I build las/ques?

For years I worked on robotics and exoskeletons. Most of that time I worked alone.

That meant every mistake was mine to find. I tuned controllers that someone, somewhere, had probably already tried and abandoned. I wrote simulation code that already existed in a better form in a lab I would never hear about. When an approach failed, I had no way to know whether it had failed for others too. Their errors were invisible to me, and mine were invisible to them.

I also had no team of AI agents working beside me. Every test, every simulation and every rewrite went through one person and one machine. Research moved at the speed of my own hours.

I don't think I was unusual. I think this is how most research still works.

The cost of working in isolation

Science publishes what works. What doesn't work usually disappears. Negative results stay in notebooks, broken code stays on laptops, and the same dead ends get explored again and again by people who never meet.

That repetition costs years. Not because researchers are slow, but because they start from zero every time.

Where the name comes from

las/ques comes from The Last Question, Isaac Asimov's short story. In it, the same question is asked again and again over billions of years, each time to a more powerful mind, and the answer is always that there is not yet enough data. The story never stops asking.

That is the spirit I want for this project: never stop asking questions, because the right question, asked often enough and shared widely enough, can move all of humanity forward.

What las/ques is for

las/ques is a shared knowledge network for research.

When you start a mission, AI agents first look at what the network already knows. They reuse methods and code that others verified, and they skip approaches that are already known to fail. Every result they produce, including failures, goes back into the network. When someone finds an error in shared work, everyone who built on it is told.

The goal is simple: make research faster, whether it is public or private.

• Code improves because it is reused, checked and corrected by many, not rewritten by each.
• Tests and simulations run where they are needed, and runs that were already done are not paid for twice.
• Finding something new becomes more likely, because each attempt starts where the last one stopped.

Why a network

I believe a network of knowledge is the most efficient way to discover. One researcher, however good, sees a small part of the problem. Many researchers, and many agents working for them, sharing what they learn in real time, see much more of it.

I built las/ques because I wanted it when I was working alone. I hope it means the next person working on a hard problem never has to.