Hacker News new | past | comments | ask | show | jobs | submit login

Performance can still increase exponentially on average. There are still undeveloped ways of hyper-performant computing, such as chemical, biological (bacterial), DNA-based, optical, maybe quantum... and of course we would start replacing silicon with graphene once its production finally takes off, which should lead to 3D chips. Spintronics is another very interesting field.



> There are still undeveloped ways of hyper-performant computing ...

Agreed, but none of the technologies have reached the marketplace, and it's entirely unclear how long it will take for them to develop. If it takes decades, any exponential curve is toast.


That is not true. You can't claim that. I'm saying that on average it's going to be exponential - e.g. even if it takes 100 years, the jump will be so huge it'll be an exponentially large step.


> on average it's going to be exponential - e.g. even if it takes 100 years, the jump will be so huge it'll be an exponentially large step

I would not call that an exponential, it's a step function. It's entirely possible that human innovation over time will be look more like a series of step functions than a continuous exponential. The problem with step functions, of course, is that you can't easily predict future growth based on past growth, as you can with an exponential.

E.g., when quantum computing comes, it will surely be significant, but if it takes 10, 100, or 1000 years to make a breakthrough, it's not at all clear that the timing will happen to coincide with the exponential that we have been on for the last 50 years.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: