Google's Head of Quantum Computing Hardware Resigns | Hacker News Reader

Google's Head of Quantum Computing Hardware Resigns

wired.com

112 pointsmmhsieh32 comments

Given the direct quote

“Since my professional goal is for someone to build a quantum computer, I think my resignation is the best course of action for everyone,”

The way I interpret this is that Quantum computing is just some pipe dream that Google isn’t taking seriously. This man intends to build a real device and either he doesn’t have faith Google will allow him to make one or has any intentions of making one.

+1 for the Google Graveyard

Marketing probably got involved, decided they should rebrand as a 'Quamtum' computer and just use standard compute.

On a more serious note, I agree with your interpretation.

I would not walk away thinking that Google is uninterested in the same goal.

Rather, there were serious differences of opinion regarding the path to get there.

Exactly. They even mentioned it in the article. Two leads disagreeing on how to get work done is never a good thing no matter where you work if they can't get to consensus.

Seems especially true for quantum computing. So many different, untested ways to get to the end goal, to the point where members of a group could all have perfectly valid reasoning as to why their particular method is the only one that will work.

In that case, if you have the resources, you find both.

You may have the money, but not the talent. I can't imagine there are that many cutting edge quantum computing researchers out there. Talk about a niche field.

LOL - isn't that the very nature of quantum computing itself?

"A quantum computer works with particles that can be in superposition. Rather than representing bits — such particles would represent qubits, which can take on the value 0, or 1, or both simultaneously" (https://plus.maths.org/content/how-does-quantum-commuting-wo...)

"The principle of quantum superposition states that if a physical system may be in one of many configurations—arrangements of particles or fields—then the most general state is a combination of all of these possibilities, where the amount in each configuration is specified by a complex number." (https://en.wikipedia.org/wiki/Quantum_superposition)

but this is quantum. you take both approaches.

That was my interpretation as well. The bizarre bit in the opening paragraph about Sundar saying that quantum computing will magically solve climate change is so baffling and stupid I'd probably want to quit after realizing that was my boss too.

I once worked as a developer for a massive trillion dollar tech company, but the company was terribly dysfunctional and never did any real work, so when I quit I joked about how I was quitting to get into software development. Sounds like the same situation here.

>when I quit I joked about how I was quitting to get into software development

I'm stealing that line.

Or, I just want to build something.

> Pichai said the milestone affirmed his belief that quantum computers might one day tackle problems like climate change

Not sure why the quote is problematic. Is modeling complex systems not a good application of quantum compute? Or is the view that climate modeling is an already solved problem?

I think the issue may lie in the logistics; can QC provide significantly better modeling (significantly in that it'll lead to significantly better outcomes) within a time frame to actually be useful? I, for one, haven't seen any QC system actually do any useful work, which is saying a lot for a multi-billion dollar "industry". It will likely be an incredibly useful technology by the end of my lifetime, but at the moment it is a marketing buzzword.

That's sort of the point, everyone just wants to create a model to further their own personal career instead of actually working on the fucking problem. I guess they just assume everything in the world will automagically fix itself once we have enough qubits?

It's either magical thinking or pure marketing department fiction, I cannot tell which.

Better models won't solve anything. It's the streetlight effect - easier to build quantum computers than change the collective behaviour.

https://en.m.wikipedia.org/wiki/Streetlight_effect

Hah. I worked at a major hospital that was a complete mess. That was basically the conversation I had with the CDO at my last job when I quit.

"I've been a data engineer for x years prior to this. I want to go back into data engineering."

My title at that job? Lead Data Engineer, of course.

The sentence doesn't have any clear meaning. That's probably why it's brilliant: each and everyone of us can and will interpret it how they like due to confirmation bias [0].

[0] https://en.wikipedia.org/wiki/Confirmation_bias

The article makes it sound more like a power struggle after Martinis got put in an advisory role which for executives is a polite way of firing someone who has done a good job.

Only half joking... guy wa probably was asked for business plan and how this will increase ad sales.

Martinis is a serious loss to them. He is a top-tier scientist that gets stuff done. He is responsible for the most successful demonstration of the power of quantum computers to date.

I suspect he wanted to continue to do the requisite science needed to build useful quantum computers, and Google began to gesture Martinis into turning it into a cloud business, following Rigetti, IBM, and Amazon. Despite the continued rapid progress, quantum computers are still too small and too noisy to solve non-toy problems, and there’s plenty of science & research to do to break that barrier of obvious utility.

If my suspicion is correct, you’ll see Google’s quantum effort turn into a marketing campaign with fancy cloud APIs that don’t serve anybody good, except bored executives who are desperate to make a mark for themselves.

I wonder how much of his team will follow him. You need someone like John when building a quantum computer and there’s not many like him.

This reminds me a bit of how disconnected IBM's marketing for Watson was from the reality the engineers saw on the ground. I wonder whether it's some sort of consequence of scale and maturity that companies fall into that pattern.

Leadership gets comfortable. Comfortable leadership has no incentive to rock the boat or strive when failure is possible, and the cushness is put at risk. Good people leave, mediocre people stay but it takes time for Rome to crumble. The cycle repeats.

Think Innovator's Dilemma [1], but at smaller scale.

[1] https://en.wikipedia.org/wiki/The_Innovator%27s_Dilemma

Martinis has always been quite honest about what their technology can and cannot do. Probably too honest for the Google managers.

John Martinis taught me physics and analog electronics! He was a great teacher: friendly, smart, and a top notch scientist to boot. All the better for UCSB that he'll have more time to be on campus.

And simultaneously doesn’t.

Why can't we use quantum computing itself to determine the best way to build a quantum computer?

We can, just need to invent time travel first.

No we don't. We can invent time travel later.

"What do we want? Time travel! When do we want it? It doesn't matter!"

tl,dr: Not now, maybe in a few years/decades.

A quantum computer is a special purpose computer. There are only a handful of algorithms known which really do better on quantum computers then on classical ones. Actually building a quantum computer is quite similar to setting up other high precision extreme matter lab physics experiments. Of course knowledge of quantum mechanics is neccessary to engineer such an experiment/computer. Evventually better quantum simulations will allow for building quantum systems which are better controllable. In the end we will probably have a similiar "self-accelerating" situation as with classical computers (which serve to develop more fancy classical computers).

Reply on news.ycombinator.com