Quantum Computing

[mattee2264]

Seems like this could be the next big thing in technology. Alphabet caught a bid recently after releasing Willow. 

 

https://www.cnbc.com/2024/12/10/google-claims-quantum-milestone-but-cant-solve-real-world-problems-.html

https://www.bbc.com/news/articles/c791ng0zvl3o

https://www.bbc.com/news/articles/c79ngx01qvro

 

I'm very much in the "not in my circle of competence" when it comes to technology but when technology stocks make up over 1/3 of the market and have accounted for the majority of the gains in the bull market that started around October 2022 it is interesting to follow from the sidelines. 

 

 

 

 

[Castanza]

If someone genuinely builds a quantum computer that is reliable, marketable, usable, and packaged in a way that would make it commercially viable; I don't see how govt would not be stepping in to hamstring the tech until there are adequate security solutions developed. Who knows what that looks like in terms of time and rollout. 

 

Pauls Security Weekly - has a few solid podcast episodes on cryptography with some experts that discuss the implications and paths forward. Worth a listen in general if you're interested in hacking, security, current trends, company issues, how companies are posturing etc. Lots of business implications discussed on the show daily by actual experts. 

 

https://podcasts.apple.com/us/podcast/pauls-security-weekly-audio/id1149992167?i=1000639854269
 

 

Edited December 11 by Castanza

[Blake Hampton]

If my understanding is correct, the environment within a computer is brought to absolute zero (-273.15 degrees Celsius) and then they start somehow turning a 1 or 0 into both a 1 and a 0.

 

This is crazy.

[Fly]

Even with recent developments it seems several years from real world application. I can't help but feel this is the latest extension of the tech rally in its final throes. What started with Mag7 ends with small cap fantasy tech. 

[Morgan]

Completely beyond my ability to understand, but it's possible this is the foundation of a new wave of computing. It will likely take a very long time before it is at scale in the world though obviously.

 

These kinds of projects from Google are a good reason not to break the company up. Almost no companies can pay for this kind of research for the years and years that it takes. Waymo is another good example. Very exciting future of course.

[lnofeisone]

5 hours ago, Fly said:

Even with recent developments it seems several years from real world application. I can't help but feel this is the latest extension of the tech rally in its final throes. What started with Mag7 ends with small cap fantasy tech. 

I like your take on mag7 to fantasy tech. I will say there are companies that have real world applications (D-wave). IBM has a great pulse on quantum world with partnerships and has made inroads with some of their clients for testing and prototype developments. 

[Spekulatius]

Quantum computer are really good at solving a few mathematical problems. They are awful at general computing. The google quantum computer solved a problem with no real application, just to show off theoretical capabilities. Any real application is at least 10 years out. 
 

[pau_]

On 12/11/2024 at 8:16 PM, Spekulatius said:

Quantum computer are really good at solving a few mathematical problems. They are awful at general computing. The google quantum computer solved a problem with no real application, just to show off theoretical capabilities. Any real application is at least 10 years out. 
 

Right we won't get a speed-up on classical computation (and really companies and state actors can scale classical computation really well with money), but the few problems they can theoretically solve are really important, and Google's announcement sounds like a breakthrough in scaling up the technology. Major state actors have been recording decades? of encrypted internet traffic with the hope that eventually they will be able to retroactively decrypt them either through vulnerabilities in the implementations or eventually through the application of quantum computers, using Shor's algorithm to break encryption that relies on prime number factorization(i.e. even if it is 10 years away it is still relevant if we think it can happen)

 

Current and future communications will be protected by a combination of post-quantum cryptography and eventually things like quantum key distribution.

 

Also applications to materials science and drug development via simulation of physical systems, but I know less about this.

[bargainman]

Anyone want to try programming one? 

https://learn.microsoft.com/en-us/azure/quantum/qsharp-overview

 

 

[pau_]

13 hours ago, bargainman said:

Anyone want to try programming one? 

https://learn.microsoft.com/en-us/azure/quantum/qsharp-overview

 

 

I have done this! I watched this lecture several years ago and built a toy example that I ran on IBM's quantum computer. If you have a decent math or CS background I think it is a very accessible intro:

 

 

 

[Spekulatius]

On 12/16/2024 at 8:22 PM, pau_ said:

Right we won't get a speed-up on classical computation (and really companies and state actors can scale classical computation really well with money), but the few problems they can theoretically solve are really important, and Google's announcement sounds like a breakthrough in scaling up the technology. Major state actors have been recording decades? of encrypted internet traffic with the hope that eventually they will be able to retroactively decrypt them either through vulnerabilities in the implementations or eventually through the application of quantum computers, using Shor's algorithm to break encryption that relies on prime number factorization(i.e. even if it is 10 years away it is still relevant if we think it can happen)

 

Current and future communications will be protected by a combination of post-quantum cryptography and eventually things like quantum key distribution.

 

Also applications to materials science and drug development via simulation of physical systems, but I know less about this.

Well quantum computing is probabilistic , so it should be good for probabilistic problems. I also think you can probably solve numerical quantum mechanical problems using quantum computing as sort of with an analog computing approach (I am guessing).

 

Also one thing to consider is that with structures getting smaller, all computers will become quantum computer weather you like it or not, because the devices become dominated by quantum effects. If you talk about structure one atom in size (3nm is 10 atoms ballpark) then an electron can tunnel and you don’t really know what is going to happen, except you will know the probability. This will probably happen in 10 years or so.

 

Interesting times.

Edited Saturday at 06:29 PM by Spekulatius

[pau_]

5 hours ago, Spekulatius said:

Well quantum computing is probabilistic , so it should be good for probabilistic problems. I also think you can probably solve numerical quantum mechanical problems using quantum computing as sort of with an analog computing approach (I am guessing).

re. numerical algorithms, if we talk about prime factorization this is where we get a huge disconnect. you basically cannot do prime factorization over any reasonably useful time horizon for breaking RSA (and you can just just pick bigger numbers as compute improves and it will continue to take a bazillion years of compute), but shor's algorithm runtime once we can implement it at this scale will operate on human-scale (hours or days for RSA with current key sizes probably? also out of my depth here) timeframes and scale with key size at a rate that grows manageably vs impossibly (^n) (the hardware still isn't there yet in terms of gates and qubits, but google's result shows promise that it is possible).

 

A good intuition to have about probabilistic computation is that it is trivial to confirm a result once you have a possible answer: you can use a classical compute step to check the results once you have them. Getting the answer is hard/impossible with classical computation, but once you have an answer, checking its validity is easy. (edit: this is not always true. I was just reading more about the willow result and it basically can't be confirmed by a classical computer at the scale they ran it; however, for things like factorization, you can just multiply the numbers together to check, trivially)

 

Agree interesting times

Edited yesterday at 12:00 AM by pau_

[Dalal.Holdings]

So is this the next bubble?

 

I guess we'll go from Crypto to Metaverse to AI and then Quantum Computing? And then Fusion Energy?

 

That's the order, right? The amusing thing is that none of these so far has fundamentally altered my day to day life, so I'm not sure where the Revenue to fund these things is coming from. But it's not like Revenue matters anymore, amirite?

[pau_]

4 hours ago, Dalal.Holdings said:

 

So is this the next bubble?

I think we're a ways off from bubble territory, but I suppose it could happen. It is more like we have a modern day Bell Labs within Google and MS and several national laboratories focused on it. I do think AI, quantum computing, and fusion will all be world-changing in my lifetime, but who knows about what companies (if any!) will benefit, and what the timeline will be. [edit: with the caveat that there is a credible idea that scale quantum computing to a useful scale is impossible due to correlated errors—there is still some scientific uncertainty for QC]

 

AI is already impacting my day-to-day in positive and negative ways. On the positive front, it is a great tool for thinking—you can't 100% trust its outputs but you can use it for brainstorming, high quality voice transcription, translation, summarization, and basic factual recall (sometimes astonishing at pattern matching) and research organization. On the negative front, I have started to see places where poor AI literacy and a hasty implementation of AI in business processes has caused people to believe things that aren't true or submit work to me that is incorrect, both in my personal and professional life. I am also seeing misinformation posts and fraudulent deepfake Musk scam ads on social media.

 

 

Edited 9 hours ago by pau_

[Dalal.Holdings]

I find myself often wasting time factchecking AI output that often turns out to be false. AI outputs currently represent midwit thinking to me at best. Furthermore, and perhaps more concerning for the firms spending tens to hundreds of billions on this, I can find very little difference from one AI platform to another.

 

If you will note, the two earlier bubbles: Crypto and Metaverse have been ultimate duds. Before that it was "people will use crypto instead of Mastercard" or "buy NFTs" or "buy real estate in the Metaverse!". No one even talks about the use case for these two things anymore. Even Zuckerberg who changed the name of his firm--what does he really have to show for it? The Raybans? An iPhone moment it was not.

 

AI also seems very overhyped and, as I noted, no one can really answer the question: where is the Revenue? What is the ROI on all this capex being spent on chips and datacenters? Of course, no one it really thinking about that and it therefore all has the makings of a bubble.

 

And the ultimate sign that AI might deflate? It sounds like Quantum Computing is the next place for speculative capital to go:

 

https://seekingalpha.com/news/4389387-as-quantum-computing-stocks-soar-wall-street-analysts-take-bullish-stance