IBM Building First Universal Quantum Computers for Business and Science | Hacker News Reader

IBM Building First Universal Quantum Computers for Business and Science

www-03.ibm.com

100 pointssocialjulio44 comments

Marketing at IBM should read up before writing illuminating sentences like "...quantum computers will deliver solutions to important problems where patterns cannot be seen because the data doesn’t exist and the possibilities that you need to explore to get to the answer are too enormous to ever be processed by classical computers."

Necessary reading (read by my high-school age cousins, so your marketers should be able to grok it) http://www.cs.virginia.edu/~robins/The_Limits_of_Quantum_Com...

I think this is one of those cases where the inverse Hanlon's Razor applies:

"Avoid attribute to stupidity that which is adequately explained by malice"

"Never attribute to malice that which is adequately explained by stupidity"

https://en.wikipedia.org/wiki/Hanlon's_razor

Has our knowledge of quantum computers changed much in the last 9 years since that Scientific American article was written? ie. Is there anything more recent worth reading? Or it's all still completely valid?

IBM's PR in AI and now this is nauseating. They are discrediting the tech industry by repeatedly promising more than they can deliver. I wonder if their PR team even speaks with their engineers, who must be cringing.

Hanhee Paik has (co-)published the following papers in 2016:

* Experimental Demonstration of a Resonator-Induced Phase Gate in a Multiqubit Circuit-QED System

* System and method for quantum information transfer between optical photons and superconductive qubits

* Rapid Driven Reset of a Qubit Readout Resonator

* Characterization of the resonator induced phase gate

* Demonstrating Multi-Qubit Operations in a Superconducting 3D circuit QED Architecture

Patents: Cavity filtered qubit

Source: https://scholar.google.de/citations?hl=de&user=lHEtpUUAAAAJ&...

This is great! There's a really cool ecosystem of Python tools for quantum computers that's coming out. Check them out if you'd like to learn more about quantum programming:

IBM's API: https://github.com/IBM/qiskit-sdk-py

Rigetti Computing's Forest: http://forest.rigetti.com/ https://github.com/rigetticomputing/pyquil

ProjectQ: http://projectq.ch/

We added a really basic introduction to quantum programming into Forest that you might enjoy: http://pyquil.readthedocs.io/en/latest/intro_to_qc.html

Fluff piece. There's no indication that they know how to build a real QC.

The part that made me certain this is more a marketing effort than a technical effort is when they described it as a followup to "Watson" and "blockchain".

"Watson" is currently a brand name for a bunch of miscellaneous NLP consulting, capitalizing on a fact that they made a very specific NLP system for answering Jeopardy questions that won on Jeopardy. The fact that the name is famous, but businesses are not actually looking to answer trivia questions, has required some marketing contortions.

"Blockchain" is, for one thing, a bunch of neat mathematics. It's also an enterprise fad that involves ignoring the neat mathematics. The logic for the fad is something like this: Bitcoin made a ton of money by decentralizing money, evading currency controls, wasting power, and selling drugs. We want to make a ton of money too! We'd rather not decentralize anything, piss off governments, increase our electric bill, or sell drugs, but maybe there's something else in there that makes money. Whatever that is, let's call it "blockchain".

I don't see a coherent category that includes Watson, blockchain, and quantum, as the author of this press release believes exists. But if I had to extrapolate:

IBM will provide an API for running a vaguely-related set of algorithms. The entire API will actually run on classical computers, but some of it will involve quantum simulators. They will leverage the confusion about where the quantum part is to sell consulting to companies where a manager has put the cart before the horse and decided they need some quantum right now.

"I don't see a coherent category that includes Watson, blockchain, and quantum, as the author of this press release believes exists."

Magical algorithms that solve everything, supposedly.

The target audience is most probably decision makers who want a business edge and not technical people.

Bitcoin is based on a blockchain process, but other applications could be as well. Blockchain exists as a separate concept. Not unreasonable to expect companies to commercialise it.

Sure, but when you take out all the parts they don't want, what they want to commercialize is "a database".

The focus is more on commercializing multi-stake-holder database replication, which is a more complicated problem than just "database" or "database replication", and is where Byzantine fault tolerant solutions apply. Hash-linked chain of blocks is just an optimization in that design space.

This is in the first paragraph:

"Quantum computers will deliver solutions to important problems where patterns cannot be seen because the data doesn’t exist ..."

It then goes on the say:

" and the possibilities that you need to explore to get to the answer are too enormous to ever be processed by classical computers"

I can understand how the second statement could be true but the first! Can someone explain how the first statement could be correct, or is the guy who wrote the sentence talking nonsense? How can you find patterns in data that you do not have?

They could be referring to simulations instead of inferring based on existing data sets.

The second statement is misleading at best. It suggests that QCs are some sort of parallel/faster classical computers.

That is (roughly) one way to look at it and actually the interpretation that David Deutsch thought brought the most clarity. He was the first to describe a quantum Turing machine.

Superimposition of marketing fluff.

Every time they don't have the data, they are superimposed with another state where they do.

How is it possible, that IBM wants to build a commercial quantum computer with 50 qubits in a _few years_, while there is a Canadian company that just released a 2048 qubit commercial quantum computer: https://en.wikipedia.org/wiki/D-Wave_Systems

I'm not an expert, please someone explain me. Is IBM's stuff is different somehow?

I'm wondering the same thing

D-Wave is not an universal quantum computer. It only performs quantum annealing: https://en.wikipedia.org/wiki/Quantum_annealing

IMB's computer should be able to execute quantum algorithms such as https://en.wikipedia.org/wiki/Shor's_algorithm to break RSA

Thanks for clarifying, I understand now.

I am pretty sure they are aware of D-Wave. There are not that many quantum computers on the market.

Surely you read the wikipedia link you posted as well :-) because it says there:

---

The study found that the D-Wave chip "produced no quantum speedup" and did not rule out the possibility in future tests.[62] The researchers, led by Matthias Troyer at the Swiss Federal Institute of Technology in Zurich, found "no quantum speedup" across the entire range of their tests, and only inconclusive results when looking at subsets of the tests.

---

There's also MSR Liqui|> that lets you program simulations in F# https://www.microsoft.com/en-us/research/project/language-in...

Recent story: Google plans to commercialize quantum technologies in 5 years:

http://www.nature.com/news/commercialize-quantum-technologie...

My limited understanding of quantum computers is that they excel at solving optimization problems, as well as problems such as integer factorization (e.g: Shor's algorithm), which is the basis for many cryptographic systems.

Now, in practice, the largest number factored using a quantum computer is very small (56,153), which is minuscule compared to the large numbers used by some standards (such as the controversial p and q numbers from Dual_EC_DRBG provided by NIST).

This seems to be the programming reference https://quantumexperience.ng.bluemix.net/qstage/#/tutorial?s... You need an account to use the editor.

Can somebody tell me something cool IBM actually does for real besides run softlayer?

I'm so tired of endless waves of marketing bullshit with no substance. At least when most companies announce something it's a tangible product I can actually expect to use at a non ambiguous point in the future

Genuine question - where do you consider softlayer to be "cool"? Or even relevant anymore - I always saw them as an "also ran" up against a big Azure and a much, much bigger AWS.

Ever heard the phrase "Nobody gets fired for buying IBM"?

IBM is massive in IT consulting with basically half their revenue coming from the consulting side of their business, and the rest they're a service company in a lot of different IT verticals.

A somewhat informative quora link, because if you were too lazy to even google this you probably won't dig deeper than this: https://www.quora.com/What-does-IBM-do

Does that change anything I said? All this marketing doesn't seem to lead to anything tangible unlike similar releases from other big companies. It's just fluff.

I'm aware the IBM does consulting and that's possibly the most boring software realm there is.

It might be boring, but it is revenue. Secondly, we should all be happy that companies like IBM (and of course Google, Facebook, Microsoft) are taking a long view and investing into cool ideas which may not necessarily immediately result in something tangible. The quarter-to-quarter view taken by most public companies is not optimal for society.

BTW, I work in ML+Medicine and IBM has dont great work in the area, with some Production implementations at hospitals.

> Ever heard the phrase "Nobody gets fired for buying IBM"?

That's now "You won't know if anybody got fired for buying an IBM quantum computer until you look inside the box".

What part of running SoftLayer into the ground is cool?

Are they? This news saddens me greatly.

Softlayer is being IBMified into nonexistence:

"In the coming days, weeks, and months, you'll start seeing 'Bluemix' more and more where you're used to seeing 'SoftLayer'. Because the legacy SoftLayer offerings and legacy Bluemix offerings will be available from a unified cloud platform, we're bringing them all under the Bluemix brand," SoftLayer's digital content director Kevin Hazard said.

"IBM isn't planning on changing the substance of SoftLayer's systems, products, services, and support, but it will phase the name out as it moves everything, including the SoftLayer blog over to Bluemix domains.

Well there's Watson [1]. It obviously performed very well in Jeopardy! and it sounds like they had some novel ideas and techniques behind the implementation.

I'm not sure how cool this is anymore however. Our company was evaluating their Bluemix [2] cloud all-the-things-as-a-service service which offers Watson. Despite their very aggressive marketing team insisting it would be able to solve a very wide range of "machine learning problems" (their words, not mine) it fell down quickly when encountering problems outside their example data-set. It was also impossible to get in touch with an engineer there, the only people you could speak to were sales and/or marketing which were useless. I get the feeling their desired clientele are technically illiterate upper-management who merely want the guarantee of a massive corporation's support and to hell with their technical capabilities.

[1] https://en.wikipedia.org/wiki/Watson_(computer) [2] https://www.ibm.com/cloud-computing/bluemix/

That is because IBM is a huge corporation, with loads of red tape. As an engineer, you are not allowed to talk directly to customers without a certain training/certification. Even if their engineers had the time and would be willing to do it, they are not allowed to.

Softlayer did have certain things that AWS doesn't (like bare metal, completely configurable VMs etc.). However, SL was a steaming pile of crap. The interface was horrible and sluggish, you'd get errors like half the time you'd order a VM, you'd get loads of errors with running VMs and many more. I don't think SL was ever a good product.

POWER is cool, even if it's probably going to die sometime in the next decade or so.

System i is cool, even if its a severely limited domain.

GPFS is really cool, even if it costs too much and isn't open source.

Cool is pretty easy to find at IBM. You just have to look where the executives aren't.

Does this mean programming paradigms will be changed? Will I have to learn a new language that's oriented towards Quantum Computing? Or is this something the computer handles regardless of the higher-level language?

Probably not, because 10-1 this is another IBM vaporware announcement that will never amount to anything.

Reply on news.ycombinator.com