IBM achieves quantum computing breakthrough | Hacker News Reader

IBM achieves quantum computing breakthrough

axios.com

26 pointspueblito23 comments

> CEO Arvind Krishna told "Axios on HBO" ahead of a planned announcement

Ah - so this is an announcement of an announcement. Those are always off topic on HN: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...

Fortunately, there's no harm in waiting: https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

The breakthrough:

"IBM says its new Eagle processor can handle 127 qubits, a measure of quantum computing power."

That seems like _alot_ Not long ago google announced a 70qubit computer

They've promised up to 1000 qubits by 2023... https://www.science.org/content/article/ibm-promises-1000-qu...

Article is a bit light on details, I'd wait until we see more results published and how they measure against other quantum computers like D-Wave's processors https://en.wikipedia.org/wiki/List_of_quantum_processors

Is this going to be true or another marketing-push that ultimately is IBM-lies? I'm thinking of Watson, the AI game changer that was thrown into the dust bin...

I am likewise having a helluva time trying to gauge just what this announcement means. Given IBM’s recent record + the CEO’s use of the word “implies”, color me suspicious.

Don't forget Blockchain - IBM was going to change the world with that.

At best it's an ongoing research project. It's wise to be skeptical of IBM statements/announcements/predictions given their track record over the last couple of decades.

Well you said it. Judge by the action not words.

Any 'breakthrough' and 'IBM' together and my BS alarm is going off.

Sounds great!!! When can we buy one from Adafruit?

Adafruit will probably have to sell cryogenic cooling equipment first.

When will PKI be broken?

8194 qubits for RSA-2048.

Does that mean 8190 qubits can't do it at all? Or it just takes longer?

Classical Shor's algorithm requires 4n+2 qubits to factor n-bits number. I just read that there are improvements which require 2n+3 qubits.

I don't think that it's possible to lower qubits number without optimizing an algorithm.

Would that mean that 1027 qbits would be enough to obtain bitcoin private keys, which is 256 bits private keys?

No. ECC is different from RSA because elliptic curve cryptography uses a more advanced mathematical basis than factoring primes.

That 256 bit ECC key is worth a 3072 bit RSA keys. At some point though people with access to quantum machines will be able to break that encryption.

Bitcoin is dead

Bitcoin?

You mean entire https web.

Financial+banking sectors would collapse? Or had to build private trusted networks? I dont even know what could be done.

Everything else - financial sector, web, etc - would move to quantum resistant PKI algorithms. It will take time and there will be issues, but that's mostly because management is dumb and incompetent and will postpone quantum resistant PKI until it's causing a lot of financial loss.

The blockchain is in particular trouble because it's decentralized. Every wallet holder would have to re-sign their transactions with quantum resistant key pairs. But this is unlikely to happen to a good number of wallets (eg Satoshi's wallet), meaning a bad actor will crack the wallets that haven't done this one by one. They will then accumulate a truly ridiculous sum of Bitcoin and devalue it for everyone else.

In fact, if someone did develop a quantum computer, they'd probably not announce it - letting it passively crack wallets, slowly accumulating vast sums of wealth, and also decrypting communications at the behest of intelligence agencies. It will likely be years after the development of such a computer that we will actually know it existed.

> The blockchain is in particular trouble because it's decentralized. Every wallet holder would have to re-sign their transactions with quantum resistant key pairs. But this is unlikely to happen to a good number of wallets (eg Satoshi's wallet), meaning a bad actor will crack the wallets that haven't done this one by one. They will then accumulate a truly ridiculous sum of Bitcoin and devalue it for everyone else.

There will be some deadline after which old wallets just will not work (essentially destroyed). So Satoshi's wallet finally will vanish.

Switch to a more resistant algorithm with a fork sign all the blocks preceding the fork through a consensus using the new method and carry on..

Bitcoin should die for many other reasons but I don’t think QC is what will kill it, I hope at some point rationality will. Crypto does represents just how broken our financial/economic system is and how easily you can push so much wealth into a non productive bubble.

>8194 qubits for RSA-2048

That's logical qubits though, right? All of these "announcements" are for physical qubits. Depending on the error correction used, it could take thousands of physical qubits just to make a single fully error-corrected logical qubit.

Reply on news.ycombinator.com