As a condensed matter physics grad student (doing scanning tunneling microscopy), I should start my reply by saying that going to grad school in physics is something that fewer people should be doing. If you want to do research in the field it is basically irreplaceable, but you have to be aware that there are many fewer spaces for postgraduate researchers, especially faculty, than there are grad students. If you are accepted at a top university, or get to work in a prestigious lab (good publications in Nature, PRL, Nature Physics, etc.), then you at least have a shot, but even then there’s not enough space and too many hopefuls. Don’t depend on everything going right, and if you have other plans, consider them. If you don’t have any other plans that are even mildly appealing, this is a warning sign that you need to spend some more time planning. A little time on plans can save you a lot of trouble.
That said, doing a PhD can force you to improve yourself. You’ll become better at doing research. It can be a lot of fun. And sometimes not so much fun, but hey, that’s why they pay you and not vice versa. Just keep in mind that if you do it, you should do it because you enjoy it, because the odds are against you being a researcher in the field in 15 years.
Okay, with the important but slightly tangential stuff out of the way, let’s talk about Moore’s law.
Why do you consider an end to Moore’s law to be bad? If you’re an unreflective computerphile, your answer might be “computers are great, and faster, cheaper computer are greater,” but I call this unreflective because it values computers based only on themselves, rather than mentioning the impact that computers have on people. If you’re a transhumanist, your answer might be “there’s people suffering and dying out there, and the faster we get to post-scarcity the better for everyone.” Or if you think AI will have a huge impact on the future, replace ‘post-scarcity’ with ‘a positive singularity.’
Around here, the typical response goes like this: the thing holding us back from a positive singularity, or a post-scarcity society where robots do the work, is not really that we don’t have good enough computers. It’s that we lack understanding. We need understanding of how to get AI to learn concepts and make plans in a complicated world, and how to train AI with complicated goal systems that capture what humans care about. If we don’t have that understanding, giving us better computers won’t help, and might be harmful if we build AIs that have a big impact but don’t care about what humans care about.
Which is all to say, I think you should not work on quantum computing if your sole motivation is the long-term impact of better computers. But I think it’s a fine thing to do if you enjoy the technical elements of research.
I think I wandered too far afield with the comment about Moore’s law (I kinda just wanted to see how people would respond, and it seemed like a more accessible question). I don’t think faster, cheaper computers are an end unto themselves, and I don’t think working on AI would be my comparative advantage . My single biggest motivation is the development of quantum simulation with a view towards quantum chemistry and many-body simulation, which in turn are relevant for the development of medicine and energy. I am a transhumanist and I want to work on technology that has a good shot of leading to an increase in life expectancies and human carrying capacity by the time it starts to really matter for me (40-50 years out, shamelessly selfish I know, I am working on becoming more altruistic. I also just enjoy thinking about physics.
I don’t want to turn this into a “predict my chances for grad school thread,” but I think I have a reasonable shot of making an impact in condensed matter research (Harvard undergraduate, I should be able to update this view after this upcoming semester when I get directly involved in experimental research on the Quantum Anomalous Hall Effect). My backup plan if academia doesnt work out is to join one of the quantum computing startups that will likely be founded over the next 7 years. I have done internships in finance and software dev, neither of which I really loved, but for a physics major/grad student those doors tend to remain at least partially open with some independent study.
I was actually not that interested in quantum computing until I came across the really beautiful idea of topological quantum computing. Also I have seen that talk by John Martinis, although it was a while ago, I am going to watch it again now that I have more of the relevant knowledge, thanks!
Part of what makes smart people valuable is that they can learn new stuff. The specialized part of an undergrad education can be done in under 2 years, 4 hours a day (even assuming no shortcuts or speedups), which is not that much time if you’re making plans for over 5 years out. So although it certainly seems like you have some strong comparative advantage in one field, you can still change tracks pretty easily.
Quantum computing startups are a bit tricky, because of the huge seed investment for low temperature and nanofabrication equipment. Maybe if people get spin qubits in diamond working at liquid nitrogen temperatures it will be cheap enough.
A great question.
As a condensed matter physics grad student (doing scanning tunneling microscopy), I should start my reply by saying that going to grad school in physics is something that fewer people should be doing. If you want to do research in the field it is basically irreplaceable, but you have to be aware that there are many fewer spaces for postgraduate researchers, especially faculty, than there are grad students. If you are accepted at a top university, or get to work in a prestigious lab (good publications in Nature, PRL, Nature Physics, etc.), then you at least have a shot, but even then there’s not enough space and too many hopefuls. Don’t depend on everything going right, and if you have other plans, consider them. If you don’t have any other plans that are even mildly appealing, this is a warning sign that you need to spend some more time planning. A little time on plans can save you a lot of trouble.
That said, doing a PhD can force you to improve yourself. You’ll become better at doing research. It can be a lot of fun. And sometimes not so much fun, but hey, that’s why they pay you and not vice versa. Just keep in mind that if you do it, you should do it because you enjoy it, because the odds are against you being a researcher in the field in 15 years.
Okay, with the important but slightly tangential stuff out of the way, let’s talk about Moore’s law.
Why do you consider an end to Moore’s law to be bad? If you’re an unreflective computerphile, your answer might be “computers are great, and faster, cheaper computer are greater,” but I call this unreflective because it values computers based only on themselves, rather than mentioning the impact that computers have on people. If you’re a transhumanist, your answer might be “there’s people suffering and dying out there, and the faster we get to post-scarcity the better for everyone.” Or if you think AI will have a huge impact on the future, replace ‘post-scarcity’ with ‘a positive singularity.’
Around here, the typical response goes like this: the thing holding us back from a positive singularity, or a post-scarcity society where robots do the work, is not really that we don’t have good enough computers. It’s that we lack understanding. We need understanding of how to get AI to learn concepts and make plans in a complicated world, and how to train AI with complicated goal systems that capture what humans care about. If we don’t have that understanding, giving us better computers won’t help, and might be harmful if we build AIs that have a big impact but don’t care about what humans care about.
Which is all to say, I think you should not work on quantum computing if your sole motivation is the long-term impact of better computers. But I think it’s a fine thing to do if you enjoy the technical elements of research.
P.S. Have you seen this talk by John Martinis?
Thank you for the detailed response!
I think I wandered too far afield with the comment about Moore’s law (I kinda just wanted to see how people would respond, and it seemed like a more accessible question). I don’t think faster, cheaper computers are an end unto themselves, and I don’t think working on AI would be my comparative advantage . My single biggest motivation is the development of quantum simulation with a view towards quantum chemistry and many-body simulation, which in turn are relevant for the development of medicine and energy. I am a transhumanist and I want to work on technology that has a good shot of leading to an increase in life expectancies and human carrying capacity by the time it starts to really matter for me (40-50 years out, shamelessly selfish I know, I am working on becoming more altruistic. I also just enjoy thinking about physics.
I don’t want to turn this into a “predict my chances for grad school thread,” but I think I have a reasonable shot of making an impact in condensed matter research (Harvard undergraduate, I should be able to update this view after this upcoming semester when I get directly involved in experimental research on the Quantum Anomalous Hall Effect). My backup plan if academia doesnt work out is to join one of the quantum computing startups that will likely be founded over the next 7 years. I have done internships in finance and software dev, neither of which I really loved, but for a physics major/grad student those doors tend to remain at least partially open with some independent study.
I was actually not that interested in quantum computing until I came across the really beautiful idea of topological quantum computing. Also I have seen that talk by John Martinis, although it was a while ago, I am going to watch it again now that I have more of the relevant knowledge, thanks!
Part of what makes smart people valuable is that they can learn new stuff. The specialized part of an undergrad education can be done in under 2 years, 4 hours a day (even assuming no shortcuts or speedups), which is not that much time if you’re making plans for over 5 years out. So although it certainly seems like you have some strong comparative advantage in one field, you can still change tracks pretty easily.
Quantum computing startups are a bit tricky, because of the huge seed investment for low temperature and nanofabrication equipment. Maybe if people get spin qubits in diamond working at liquid nitrogen temperatures it will be cheap enough.
But, hey, if it floats your boat, go for it.
P.S. your link should go here: https://www.youtube.com/watch?v=igPXzKjqrNg