Hacker News new | ask | show | jobs
by AlexNicita 42 days ago
Does contemporary technology lack dimension?

Picture perfect symmetry between problems and solutions. Or imagine a machine isomorphic to all the puzzles it solves. Why is it that so few tools accomplish this?

The behavior of complex dynamical systems is nuanced to model; visualizing the shape of technology is an even more elusive idea. Paul Graham recently wrote in the shape of essays1 that with good writing, "There's a continuum of surprise." Words can create the sense of magical discovery, where learning unexpectedly valuable knowledge is truly surprising. Beautiful technology presented to its user evokes this same exact feeling, essentially unconscious joy, and if that unique feeling were to have shape, it would be a continuum. Such technology would be unbounded, limitless, and infinite: n-dimensional.

This is easier to say than to build, so rooting the metaphor in reality is more illustrative. In my opinion, the modern technologies that are closest to being n-dimensional are in the emerging field of automated software engineering. The field is advancing rapidly. Cursor is going for $60B as an option. Codex is accelerating every possible aspect of intellectual processes. Cognition is executing well on enterprise. These are all historic moments that obsolesce SaaS as we know it, and the market is just getting started.

It's a worthwhile puzzle to predict the future of n-dimensional machines. Traditional operating systems were the n-dimensional machines of their time and that market had many winners. Observing the growth of foundation models as products, it is reasonable to expect the same from the major labs. Anticipating the next cycle of this evolution is perhaps the most interesting open question in business today. The answer must lie in platforms built atop models, businesses wrapping wrappers, and creating newly infinite machines from the ones that today seem limitless. Inevitably, today will cede to future supercycles. Or the software development lifecycle will have to altogether evolve.

Personally, I believe that cheaper, simpler, and longer horizon laissez faire technology will be among the most critical n-dimensional machines for enterprise and consumer, comparable to a robotic tractor tending to crops on autopilot (for software engineering). Rote, repeatable abstractions will need to be automated across "vibe-coded" software. These layers are slowly becoming obvious—dependency upgrades, style consistency, and even simple user feedback—all are becoming automatable iterations sans coder. https://constantcoder.ai is an early sketch of one such future n-dimensional machine where the system design works to obsolesce prompting, plus most other human input.

Onward, toward technology of true breadth and depth; toward n-dimensional machines.

Godspeed.