This is all very impressive, kudos to Google and all, but unless QEC demonstrates stable multi-qubit entanglement - stable enough to execute operations needed for Shor’s algorithm, for example - all that machinery accounts for nothing. Having qubit in superposition state just gives you 0 and 1 and confers no special benefit, only entanglement between qubits does. Until they explicitly address collapse of then entanglement and demonstrate its extended lifetime, their demonstration is smoke and mirrors. QC explicitly benefits from a narrow range of algorithms that require entanglement, the rest can be accomplished by regular computing with a fraction of cost.
This is true. But that's the nature of all experimental research. There have been lots of scientific discoveries that only really found value after decades.
I think work like this can help us push boundaries b/c it leads to new avenues of research that lead to breakthroughs that we would've never considered. A company like Google has the profits to invest into projects like this, which I would prefer infinitely over stock buybacks or spending it on regulatory capture etc.
Yes, it with QC - there are very specific algorithms that benefit from the quantum nature of qubits, the rest - classic calculations - have no benefit. It’s fundamental and I don’t think there is disagreement with this statement. They can push boundaries for sure, but at the same time they provide these market projections for the hardware while the foundational part - qubit entanglement - is not present. This is troubling. QC has been at this for 20 years now and I honestly see them selling magic blue water to investors.
Firstly, wrt to projections- I agree to an extent. I think a lot of these are based on investor speculations, based on the assumption that QC will eventually return. It's essentially betting that if QC becomes 100 years from now, then what would the value have been at this stage. We have no way of knowing that, but that's research.
Complex numbers existed for centuries before we had practical use for them. Now they're integral to us. Similarly true for many of Computer Science and Scientific Principles.
As we covered in the Bell Labs piece, cutting edge research is hard b/c we often don't know what we don't know. I'm not saying QC itself will be successful. I'm saying that most of what we do will likely fail, but we do it anyway b/c what succeeds will be game-changing (kind of the VC model but science).
QC fails commercially but leads to tech that is valuable (NASA inventing great science going to space).
QC succeeds but no returns
QC succeeds and massive returns.
Imo the outcomes of possible returns of 2 and 4 outweigh the risks.
You could talk about opportunity costs and whether this is the way to utilize the funding vs other tasks. And I'd probably agree. But that would also involve a larger discussion of reworking the economic system.
While I agree with overall premise that it’s hard to predict the value of fundamental research and its fruits can be unexpected - with QC I think we have 2 fundamental assumptions: its usefulness comes through QC algorithms and those algorithms rely on mutti-bit entanglement.
Without entanglement, we have 2 states per qubit and there are no special benefits in operating just on them in known algorithms,
Multi-qubit entanglement is hard with photons or trapped ions - and probably impossible with solid-state qubits. If the QC research does not address either algorithms (finding new ones) or qubit entanglement, they deal with secondary objectives which - in my opinion- do not advance QC but rather add smoke and mirrors to already complicated problem. Sometimes I suspect all these QC researchers just laugh quietly after having a sip of whiskey in their lab suspecting full well that this is just a well funded perpetuum mobile project.
You might end up being correct. I see this research like Complex Numbers (which suddenly came into relevance after centuries of being useless). But it might end up being a dud.
Assuming you're correct- where would you redirect the money, under the constraints of our current economic system?
Could I do better than DARPA? Probably not. CHIPS Act just is distributing unholy amount of bacon grease - mostly on advanced packaging as far as I know. We hopefully will see in 5-6 years if those bets pan out. But for QC - of course, Google can do with their cash whatever they want - algorithms and entanglement are the hardest part, especially the latter.
In the short term, this gets bought by governments and other organizations that are researching quantum computing. The best part about this is that you can buy things like Quantum Computing Data (or you will be soon)+ Synthetic and build for relatively cheap now. This leads to better QEC groups which can enable more Quantum Computing.
This is the best article to read to gain an understanding of Google's Willow chip and what that means. It provides a very realistic perspective on quantum computing grounded in actual research and explained very clearly.
Just to clarify- this not on Willow, since we don't know the details there. I'm fairly sure that is this is related since this talks about EC, which is a huge key to Willow. But we don't know if this is what they for sure
Devansh,
This is all very impressive, kudos to Google and all, but unless QEC demonstrates stable multi-qubit entanglement - stable enough to execute operations needed for Shor’s algorithm, for example - all that machinery accounts for nothing. Having qubit in superposition state just gives you 0 and 1 and confers no special benefit, only entanglement between qubits does. Until they explicitly address collapse of then entanglement and demonstrate its extended lifetime, their demonstration is smoke and mirrors. QC explicitly benefits from a narrow range of algorithms that require entanglement, the rest can be accomplished by regular computing with a fraction of cost.
This is true. But that's the nature of all experimental research. There have been lots of scientific discoveries that only really found value after decades.
I think work like this can help us push boundaries b/c it leads to new avenues of research that lead to breakthroughs that we would've never considered. A company like Google has the profits to invest into projects like this, which I would prefer infinitely over stock buybacks or spending it on regulatory capture etc.
Yes, it with QC - there are very specific algorithms that benefit from the quantum nature of qubits, the rest - classic calculations - have no benefit. It’s fundamental and I don’t think there is disagreement with this statement. They can push boundaries for sure, but at the same time they provide these market projections for the hardware while the foundational part - qubit entanglement - is not present. This is troubling. QC has been at this for 20 years now and I honestly see them selling magic blue water to investors.
Firstly, wrt to projections- I agree to an extent. I think a lot of these are based on investor speculations, based on the assumption that QC will eventually return. It's essentially betting that if QC becomes 100 years from now, then what would the value have been at this stage. We have no way of knowing that, but that's research.
Complex numbers existed for centuries before we had practical use for them. Now they're integral to us. Similarly true for many of Computer Science and Scientific Principles.
As we covered in the Bell Labs piece, cutting edge research is hard b/c we often don't know what we don't know. I'm not saying QC itself will be successful. I'm saying that most of what we do will likely fail, but we do it anyway b/c what succeeds will be game-changing (kind of the VC model but science).
Bell Labs- https://artificialintelligencemadesimple.substack.com/p/what-allowed-bell-labs-to-invent
The way I see it there are 4 options-
QC fails and was a giant waste of time.
QC fails commercially but leads to tech that is valuable (NASA inventing great science going to space).
QC succeeds but no returns
QC succeeds and massive returns.
Imo the outcomes of possible returns of 2 and 4 outweigh the risks.
You could talk about opportunity costs and whether this is the way to utilize the funding vs other tasks. And I'd probably agree. But that would also involve a larger discussion of reworking the economic system.
While I agree with overall premise that it’s hard to predict the value of fundamental research and its fruits can be unexpected - with QC I think we have 2 fundamental assumptions: its usefulness comes through QC algorithms and those algorithms rely on mutti-bit entanglement.
Without entanglement, we have 2 states per qubit and there are no special benefits in operating just on them in known algorithms,
Multi-qubit entanglement is hard with photons or trapped ions - and probably impossible with solid-state qubits. If the QC research does not address either algorithms (finding new ones) or qubit entanglement, they deal with secondary objectives which - in my opinion- do not advance QC but rather add smoke and mirrors to already complicated problem. Sometimes I suspect all these QC researchers just laugh quietly after having a sip of whiskey in their lab suspecting full well that this is just a well funded perpetuum mobile project.
You might end up being correct. I see this research like Complex Numbers (which suddenly came into relevance after centuries of being useless). But it might end up being a dud.
Assuming you're correct- where would you redirect the money, under the constraints of our current economic system?
Could I do better than DARPA? Probably not. CHIPS Act just is distributing unholy amount of bacon grease - mostly on advanced packaging as far as I know. We hopefully will see in 5-6 years if those bets pan out. But for QC - of course, Google can do with their cash whatever they want - algorithms and entanglement are the hardest part, especially the latter.
Opinions on the initial practical applications of this? (From a market perspective - i.e. what problems will people pay to solve?)
Supply chain, stock market, evolutionary biology, quantum chemistry :)
People are still far away from practical applications, even in quantum chemistry.
Practical Applications are far away, but I have a lot hope
How exactly QC would do these better than regular computers? What algorithms would take advantage in these areas?
In the short term, this gets bought by governments and other organizations that are researching quantum computing. The best part about this is that you can buy things like Quantum Computing Data (or you will be soon)+ Synthetic and build for relatively cheap now. This leads to better QEC groups which can enable more Quantum Computing.
This is the best article to read to gain an understanding of Google's Willow chip and what that means. It provides a very realistic perspective on quantum computing grounded in actual research and explained very clearly.
Incredible writing, Devansh!
"the best"- Pretty sure I'm the only article on this lmaoo.
But srsly, thank you Logan. Glad you liked it.
On this, yes. On Willow, no. But I think this is more beneficial to understand Willow than articles focused on it.
Just to clarify- this not on Willow, since we don't know the details there. I'm fairly sure that is this is related since this talks about EC, which is a huge key to Willow. But we don't know if this is what they for sure
This definitely shows why Willow is important.
Thanks for a great article!
Thank you Hugo.
saved for weekend
Lmk what you think!
Great work Devansh
<3
Great piece, thanks for putting this together!
<3