The story of how Cray and his team created the CDC 6600 is really great [0]. Cray decided they had to get away from corporate interference so he picked a bucolic town a couple hours away from corporate HQ, moved his whole team there, and they worked in peace to develop a machine that was 10x faster than anything else.
It's unclear today if quantum computers will be useful, just as it was unclear in 1962 if supercomputers for large scientific numerical calculations would be useful. It'll be interesting to see.
For anybody interested in quantum computers I very enthusiastically recommend taking a look at Scott Aaranson's blog - http://www.scottaaronson.com/blog/
I don't think quantum computing will have any sort of impact on drug discovery. Chad mentioned quantum computing's impact on drugs. Anybody know what he's referring to?
I haven't seen anything to suggest that there's big problems out there waiting to be solved by powerful computing capabilities. Has anyone else found something that suggests there's active problems waiting for quantum computers?
I've always found the missing thing in life saving pharmaceuticals is sick people to test compounds on.
Nonetheless, good to see YC pushing the envelope on impactful stuff.
I hope they desperately bring some inorganic and organic chemists with 10+ years of experience. It would open a new can of ventures that are meaningful. The good problems, the great ones, are in chemistry.
He means using computational chemistry to predict what different drugs might do without actually having to produce them one by one first. We are terrible at predicting how large molecules will act because they are complex quantum systems. So using a quantum computer to model them will hopefully be very successful.
> I don't think quantum computing will have any sort of impact on drug discovery.
I also was very surprised by that. I think it's probably 98% hype and 2% truth. There might be cases where the better understanding of physics brought by QC will impact a better understanding of biology. But I'd be very surprised if that made a big impact.
> I've always found the missing thing in life saving pharmaceuticals is sick people to test compounds on.
I don't think that's the case either. I think people who have a life-threatening illness are pretty willing to take part in trials.
What I'd rather think is that likely the most important thing is to apply stricter scientific standards. Pharma companies have been complaining that they can't replicate the vast majority of pre-clinical cancer studies. The pharma companies themselves are also guilty of skewing research by hiding results and not sharing data.
That's the thing that's holding up research. There's too much crappy science that wastes resources.
The main computationally intractable problems around drug discovery surround protein folding and identifying macromolecule shapes. This is necessary to infer receptor sites for potential drug targets.
The problem is massively intractable. QC could revolutionize the space.
Remember though, NP complete problems still are intractable (exponential complexity) even for quantum computers, to say nothing of NP hard.
Unger and Moult (1993) have shown a three-dimensional protein
folding model to be NP-complete, and a two- and three-dimensional
mathematical model describing the folding process as a free energy minimization problem is NP-hard:
I'm curious what the MVP is? Despite the whole counterargument to Eric Ries school of thought, even YC's thesis is that "hard companies" should have some sort of initial stepping stone.
I'm curious why he pronounces Regina incorrectly (he went to the University of Regina), when he is from Moose Jaw, a 50 minute drive away. It's really really strange. 7 year old children all across Canada know how to pronounce that city correctly...how does a guy forget something so basic?
15 comments
[ 117 ms ] story [ 2685 ms ] threadIt's unclear today if quantum computers will be useful, just as it was unclear in 1962 if supercomputers for large scientific numerical calculations would be useful. It'll be interesting to see.
[0] https://www.amazon.com/Supermen-Seymour-Technical-Wizards-Su...
"Never doubt that a small group of thoughtful, committed citizens can change the world; indeed, it's the only thing that ever has."
https://www.youtube.com/watch?v=43DOgo2VzGU
I haven't seen anything to suggest that there's big problems out there waiting to be solved by powerful computing capabilities. Has anyone else found something that suggests there's active problems waiting for quantum computers?
I've always found the missing thing in life saving pharmaceuticals is sick people to test compounds on.
Nonetheless, good to see YC pushing the envelope on impactful stuff.
I hope they desperately bring some inorganic and organic chemists with 10+ years of experience. It would open a new can of ventures that are meaningful. The good problems, the great ones, are in chemistry.
Biology simply isn't that discrete. It's like 4000 years of spaghetti code.
I also was very surprised by that. I think it's probably 98% hype and 2% truth. There might be cases where the better understanding of physics brought by QC will impact a better understanding of biology. But I'd be very surprised if that made a big impact.
> I've always found the missing thing in life saving pharmaceuticals is sick people to test compounds on.
I don't think that's the case either. I think people who have a life-threatening illness are pretty willing to take part in trials. What I'd rather think is that likely the most important thing is to apply stricter scientific standards. Pharma companies have been complaining that they can't replicate the vast majority of pre-clinical cancer studies. The pharma companies themselves are also guilty of skewing research by hiding results and not sharing data. That's the thing that's holding up research. There's too much crappy science that wastes resources.
Finding someone with the perfect storm of conditions is really really tough.
The problem is that all the studies really have to be n=1 and the hunt for drug targets is now the hunt for the black swan.
The problem is massively intractable. QC could revolutionize the space.
Unger and Moult (1993) have shown a three-dimensional protein folding model to be NP-complete, and a two- and three-dimensional mathematical model describing the folding process as a free energy minimization problem is NP-hard:
https://www.gwern.net/docs/1993-fraenkel.pdf