Hacker News new | past | comments | ask | show | jobs | submit login
Delft scientists make first “on demand” entanglement link (tudelft.nl)
183 points by amelius 9 months ago | hide | past | web | favorite | 73 comments

"Entanglement forms the link that will provide a future quantum internet" -- gee, if it was hard to explain to people that entanglement doesn't break the speed limit by transmitting information instantaneously before.. yeah we're going to need a simple go-to explanation of what this actually is and how it works.

Here's your simple go-to explanation: The quantum internet is a buzzword to extract research money from EU programs. It has no relevance whatsoever for the real internet.

Considering the article is from the press office it is full of buzzword bingo words and meant to be that way. The real article (haven't found a preprint yet) should be far more scientific and worthy of notice. The 'internet' won't benefit directly, today, but might somehow in the future.

As with all fundamental research: we don't know yet. Let's not be judgemental beforehand.

I was under the impression that quantum entanglement could provide perfect privacy across a connection, as it would be obvious whether the line communication was intercepted.

This is the impression they want you to have. There are a few caveats:

* You need a direct connection with a limit in length. This means effectively you can't have wifi, mobile internet or anything like that. You also probably can't use it to have a connection between the US and Europe. Calling that a "quantum internet" is... only possible if you have no idea what the term "internet" means.

* You need an existing authenticated channel or be vulnerable to man in the middle attacks. The quantum key distribution schemes always assume this already exists, but never care to explain where they got it from.

>You need an existing authenticated channel or be vulnerable to man in the middle attacks

Isn't this theoretically unavoidable for all forms of encryption? You need some way of distinguishing your intended communication partner, because the best that any protocol could do is establish a secure link with whoever is on the other end of the line.

From what I understand, quantum entanglement is used in "secure communications" as a key exchange (more of a distribution, as a third-party is required to distribute the entangled photons) (QKD - See: BB84 for example). Following this key exchange phase, normal encryption algorithms are used over traditional infrastructure. This is already being done in real environments.

In a real world implementation, you are using this as a key stream, of which interceptions can be detected by certain loss characteristics.

The problem is that normal encryption is complicated and often has vulnerabilities.

Perhaps we could find a way to transport an entangled photon down some channel?

But you'd need an entangled pair with everyone you want to communicate with.

You'd just have to organize daily deliveries from the Entanglement Man.

Chances are the Entanglement Man makes 12.00/hr, has to provide his own car and pees into a bottle to make quota.

No, you would just need to get an entangled key from a distributor. Once you can communicate with one well connected and trusted actor, you're good to go.

Think more like web security certificates...

But how do you know you have the right entagled peer? How's that different than any other out of band key exchange?

That's true for the E91 protocol, but for BB84 you only need a (public) quantum channel through which Alice can send states to Bob.

Or, more generically, whenever adding for government grants, use the buzzwords of the day

Today it could be block chain, yesterday it was telemedicine, etc

You forgot to mention Quantum Cloud Computing. The first killer feature of the new and fantastic Quantum Internet.

I think you mean Quantumcurrency. Guaranteed to be cryptographically secure, and you'll only know if you own some by collapsing the waveform.

EU programs are forcibly funded by a tax payer, paying obscene amount of tax. I have 0 respect to companies participating in such programs.

Wouldn't light-speed internet be useful ? I would think that HFT firms would be interested in them

We already have "light-speed internet" or close enough. Signals in cables travel at (approximately) light speed (for the medium, which is not vacuum). And we do have wireless EM signals and fiber optics cables as well...

The velocity factor is around 2/3 in optical fiber. That's slow to HFT firms. For example, the delay compared to vacuum/c would be of of 2ms for a 1000km link.

And that does not take into account fiber repeaters or microwave links

You’d basically need a 1000km vacuum chamber to achieve that, which would be insanely expensive and an engineering miracle. I’m not even getting into what you’d be using as a transmission medium, because the vacuum chamber is prohibitive enough.

No, there is hollow fiber for communication. https://gizmodo.com/5992303/hollow-fiber-optic-tunnels-can-b...

but also reasonably low data loss of 3.5 dB/km. That still adds up at a distance though, so these crazy fast cables are most likely destined for supercomputer and data center applications, for now at least.

Not really for communication, and I suspect that if you want it to scale up to 1000km, it’s going to strongly resemble the vacuum chamber I mentioned.

You could use LEO satellites, add a few km up and down but you have way less hardware to pass through.

That’s a good point, the difference between vacuum and atmosphere isn’t so significant, but then it’s not really a new tech or “lightspeed internet” it’s just a satellite link.

With the addition of satellite-to-satellite optical connections, which will greatly improve latency and bandwidth.

If SpaceX pulls off their swarm of telecom satellites, we might get to see how it really works in widespread practice. Exciting prospect!

Indeed. And what I'm most interested in is all the cash they're dumping into commercial phased-array antennas. If they manage to scale enough to get them to a low enough price they will be amazing for all sorts of other stuff.

>That's slow to HFT firms.

That's because we allow them to trade at such rates and compete on speed.

A mere few decades to half a century ago we have 1/10 to 1/1000 the speed, and trading was going on just fine...

I'd opt for a "turn based" trading system (with a small, but not tiny turn window), to eliminate any transmission speedup advantage from different traders.

If they're the only people interested in that, I'm okay with staying at 2/3...

Yeah, what the hell happened to the no-communication theorem? I'm guessing bad science journalism inserted this quanternet idea in a fit of zeal?

If it does happen and it gets named The Quanternet, I want some credit by the way. Y'all heard it here first.

Typically (thinking back to uni days) these things are done more for the (statistical) guarantee that nobody has peeked at what's going down the wire.

Though I expect changes like this are probably more broadly applicable.

Please provide a brief one, or a few links! I'd love to know.

You can't choose what is the value you read from an entangled particle, thus you can't choose what value your peer reads either.

Your peer will read the inverse value that you read, but that alone does not make any information go through the channel.

Oh it's better than that! The term "quantum teleportation" is horribly misleading but that would be a major application of this. What "quantum teleportation" should maybe be thought of is allowing you transmit a quantum bit to someone else by instead (A) sharing an entangled particle pair ahead of time and (B) transmitting two (non-quantum) bits to them over, say, ethernet. If you prepare your half of the entagled particle pair properly and send them the right two (classical) bits, then they will be able to reconstruct your original quantum bit out of their entangled particle. So it's a way of getting my quantum bit to your quantum computer without having to pick it up and move it over there.

It doesn't happen faster than the speed of light since it still involves transferring 2 regular bits. And it's not really teleportation, unless I "teleport" an image to you by sending emailing it to you. The main difference with sending regular bits is that you inherently destroy your original quantum bit before you "teleport" it, which is presumably where the name comes from. Since copying quantum bits is impossible (the "no-clone theorem"), you can't work around that destruction either.

I know, like, nothing about this. But my wiki reading makes me think this is essentially a one time pad. you share the entangled bits, you send a message, the message tells you what to do if your entangled bit is one, or zero.

nobody can mess with the entangled bits, or they lose entanglement, and the message becomes random.

i dunno, it seems like if you had a bunch of them you could make a secure channel.

Quantum teleportation is in fact a quantum generalization of the classical one time pad. And it does form a secure channel, but with the same caveats as in the classical case (e.g. it's not authenticated).

Many quantum things have classical analogues. For example, there is classical no cloning [1] and classical delayed erasure [2], though people will debate if these classical versions have really captured the "essence" of the quantum things.

1: https://marozols.wordpress.com/2012/05/21/three-myths-about-...

2: http://algassert.com/post/1720

> i dunno, it seems like if you had a bunch of them you could make a secure channel.

I don't think anybody solved man in the middle attacks against it yet. If the attacker takes hold of both the entangled channel and the classical channel, he can impersonate both parties and keep the conversation going.

And you can't store those pairs. All those articles are about people that made them survive long enough for transmission. So you can't distribute them in person.

That MITM attack is considerably more fragile than just reading a one-time pad during transit, though.

Asymmetric encryption is cheaper, survives storage, requires only the transmission of public data, and is not fundamentally flawed like that. It is also more versatile.

But non-quantum encryption hasn't been mathematically proven to be secure. In the real world, you'll probably always use both and only if large quantum computers become available.

Quantum computers are not needed for quantum encryption, and quantum computers do not break classical encryption in general, just the most used algorithms. They simply don't change the overall picture on this.

Yes, non-quantum encryption has not been proven secure. That beats a known insecure systems at any time. You are also underestimating the cost of quantum crypto - it will only work on direct fiber, it can never be routed.

Maybe some classical algorithm is immune to quantum computers, but none have been proven to be, unlike quantum encryption.

Also, the Chinese have already demonstrated quantum channels over satellites - it doesn’t need fiber. The satellites are certainly acting as routers too.

Quantum encryption is broken today. Classical encryption may be broken, nobody knows.

None of those is perfect, but one is better than the other.

Could you please elaborate?

If I understand correctly what is written here, observingicle causes it to increase its, and its peers, state. Wouldn't this let you to decide afterwards if you wanted to again observe it and invert the state if you wanted? Or does observing break entanglement / an inverted particle doesn't get inverted back after another observation?

Entanglement is broken on observation. “Mike and Ike” suggest we may have a “department of entanglement“ someday like we have a DOE. It’s a scarce resource.

About the best application I’ve read so far is using entanglement as a signal amplifier (my over simplification).

But if you have n entanglements you can get n bits of information from the channel by timing the disentanglements at agreed-upon intervals, no? Like, Morse code...

You can't observe the entanglement breaking. You can only observe the particle's state.

>If it was hard to explain to people that entanglement doesn't break the speed limit by transmitting information instantaneously before..

Blame whomever came up with the term "quantum teleportation" for what should probably more correctly be called "quantum transmission."

Wouldn't a "quantum internet" have more stable latency? I generally notice that my latency fluctuates between 50ms and 100ms. I attribute this to the other users on my ISP's network, all the routers/switches/etc between me and the server I'm connecting to, and the load the server is under. I assume that the first two things (users and equipment) would be mitigated by using entanglement to communicate. Can someone tell me why this line of reasoning might be wrong?

You wouldn't have point to point links with everyone else. You'll still have common trunk lines that get congested.

Does there even exist a working quantum network at the moment, as in has anybody ever connected two devices and sent 1MB of data over the link (even if the 2 nodes are only a few meters apart)?

This article is talking about building an internet, but I'm not even aware of a reliably functioning peer-to-peer network, which is surely a needed first step?

I'm sorry to sound like the idiot here but, does this mean much lower latency, if so by how much?

Furthermore, how does this make communication more secure? Are you not still sending data from node to node?

Information never goes faster than light, even though quantum information does. But, the communication is more secure because the data which goes from node to node is quantum encoded. So if someone would eavesdrop, the code would break and the receiver would immediately know there was an eavesdropper, as he would no longer be able to decode the information. Quantum information can only have 1 sender and 1 receiver.

It reduces the latency a little bit as you only have the distance between two points, rather than the longer path that it would take for a signal to travel in a cable.

In addition the speed of light is on the order of 30% slower traveling through a fiber optic cable than through a vacuum. This probably will be where the biggest gain in latency will be.

No. Information doesn't travel faster than light, ever.

Hehe. Offer he can't refuse. Lol.

In layman terms does this mean we will get 0ms latency?

Not at all. Instantaneous communication can be used to send messages back in time, so it would be a much bigger deal if that was expected.

Can you expand on this? How can it be used to send messages back in time if both are timestamped to the micro/nano second?

I believe it is a current law of physics that even the propagation of information is limited to the speed of light. So, no.

Well, laws can change. Both legal laws and laws of physics. There have been several laws of physics in past centuries that have been found not to hold (not that it's very probable, but still possible).

Ha I don't think causality is going to change. That's a lot different from all those other times science changed. It's not only the underpinning of all physics, confirmed by literally every single scientific observation in the history of mankind--it's also just logically inconsistent to break causality. It makes no sense at all. Causes happen before effects, not ever the other way around.

I guess maybe we could someday find an exception to causality, but it would be a way bigger deal than any other discovery we've made. It would be a back to square one situation.

As they say, you can have FTL, relativity, or causality--pick two. We've really nailed down relativity and causality. It's not looking great for FTL. Trust me, it breaks my heart every time I think about it. But that's the way it is sometimes.

>it's also just logically inconsistent to break causality

Half of QM (or its interpretations) is also "logically inconsistent", though.

>Causes happen before effects, not ever the other way around.

That's from our own perspective though. In certain theories of the universe (even ones consistent with relativity: https://en.wikipedia.org/wiki/G%C3%B6del_metric) time is not the one way stream we conceive it to be.

Interpretations have nothing to do with logical inconsistency. They've got even less to do with the Theory itself. You, on the other hand - are putting interpretations and the theories in one bucket.

Godel's metric have nothing to do with reality if you actually read about it's properties. For example, the universe described by it would have no reshift for distant objects.

Please, check your facts. Otherwise you sound just like an uneducated new-ager looking for "truth".

FWIW my wording was _current_ law.

Off topic: I was in Delft recently on a whim when visiting a friend in Rotterdam. Beautiful old town, highly recommend passing through. And only 10 mins by train from Rotterdam!

If you want to visit The Netherlands don't (only) go to Amsterdam. It's like only visiting Disneyland when you go to the US. It's really nice but you don't really get a realistic picture of the country (at least the center).

Madurodam was a fascinating day visit. Kidd-ish, but some of the models were fascinating.

Advice: don't drive a car in Amsterdam! :)

Yup, and a good tech campus as well

I can recommend taking one of the water tours. It's a really cool place, if I had to live anywhere in the west of the country it would be my top choice.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact