Hacker News new | past | comments | ask | show | jobs | submit | lxgr's comments login

Some hydroelectric plants can pump water back up and both absorb excess supply and provide stabilization during demand spikes.

I have no idea if it’s feasible to fill longer-term gaps during extended cloudy/windless days and nights using that alone, though. Other than that, there are already large-scale battery plants deployed in some cities.

Another approach is to control the demand side: When air conditioning or heating with electricity, minutes usually don’t matter, and dropping/providing extra load at very short notice should be feasible in a smart grid.


It is entirely infeasible to used pumped hydro alone to solve our energy storage problems. They’re a great component of the solution but they’re big expensive environmentally disruptive projects and there are limited suitable sites.

They're also quite inefficient (75% ish, which is not horrible, but still something) as you lose energy both in pumping and generation.

So the more you use them, the more energy you need in the first place, so you try not to use them too much. And the worst thing for a capital intensive business is to not get used much. Even with the arbitrage advantage, it takes a long time to pay it off. The grid may pay a "retainer" to sweeten the deal but you can't just build more and expect them all to get that benefit.

It's similar to the problem that "use excess energy to electrolyse water into hydrogen" has: no-one running a multi-billion electrolyser really wants to run it a few hours a day only when the grid is oversupplied. And in top of that, risk being cut off at the knees decades before breakeven if someone comes along with a cheaper/more profitable way to deal with the oversupply.


Yes, which makes the point of dedicated connectivity moot from a data security point of view.

Metadata security and availability are different concerns.


>Metadata security

If you have a dedicated circuit, you can send dummy data 24/7 to mitigate any traffic analysis. Even if you don't, you configure each link to send dummy data, so eavesdroppers can't do any traffic analysis without compromising the node itself.


Traffic analysis tells your adversaries who you communicate with, and what apps you're using, inferring what communication was caused by what preceding communication, etc which lets your adversary guess what the communication was about. Esp when compared against what other people is communicating about just then.

Yes, which makes it a particularly bad idea to run unencrypted metadata over satellite connections if it can be avoided – and that's the case for communication between data centers, arguably.

That might change once lasers or extremely tight radio beams can be used for ground stations, but for the latter you'd still need to make sure that nobody can get reasonably close to your ground stations, which might be possible for remote military bases, but probably not for AWS data centers.


> This enables secure private communication outside of the internet across data centers

How so? I'd imagine the datacenter terminal side downlink to be much more easily tappable than fiberoptics.

There are advantages in latency and potentially availability, but even there I would imagine fiber to win in an adversarial active jamming scenario.


> easily tappable

I suppose in any realistic scenario we should assume that the enemy may be listening to all our communication at all times. This is the assumption behind such daily things as WPA3, SSH, TLS.

Jamming is a much more serious concern.


Yes, so all in all for satellite vs. fiber in backbone applications, I'd say that it's a wash (or slight win for fiber) when it comes to security, and a definitive win for fiber when it comes to jamming resistance.

In the field it's a completely different story, of course – you can't always pull fiber (although it does appear in unexpected scenarios, such as fiber-operated UAVs or torpedoes).


Wire-guided drones and missiles seem to be increasingly common, probably due to the cost of instantaneous radio jamming being so low - that's a short-lived signal that is going to a vehicle near your adversary. Torpedoes make a lot of sense for other reasons because radio has a very hard time getting through water (and things like sonar have too low bandwidth as well as giving away your position).

However, it's very easy to cut a fiber in a way that is hard to repair. Fishing trawlers do this all the time. In that sense, fiber can be "jammed" (sabotaged) much more easily than radio/satellite.


Not sure. Jamming a wireless signal is an active process that stops interfering when you stop transmitting.

Destroying fiber with a backhoe or an axe doesn't stop interfering when you stop digging or chopping though.


FHSS[1] has made jamming difficult in US military communications for decades. It doesn’t make it impossible but jamming the entire spectrum is nearly impossible at scale for almost everyone. At best it would affect small areas until the US sent rf seeking missiles (HARM are designed for this) at the jammer source. Also note that modern satcom like Starlink uses AESA digital phased array antennas much like a F35’s radar. It’s so much more complex than legacy analog stuff.

[1] https://en.m.wikipedia.org/wiki/Frequency-hopping_spread_spe...


It’s easily possible to jam a frequency spread signal from a satellite, assuming you can get the directionality right (i.e. you need to be in the beam of whoever you’re trying to jam, or you’ll need even more power to overcome their receive directionality, which is never perfect).

Signal strength (satellites are power constrained) and distances involved are tough.

GPS uses frequency spreading too, and locally jamming that (even the military version with a secret/unpredictable spreading code) is trivial, for example.


US military gps is encrypted and heavily used FHSS. It is jam resistant, and can’t be jammed without significant expertise and know how.

Russia has some of the best EW chops in the world (after the NSA perhaps), and they struggled to successfully jam Starlink after some defensive work was carried out by SpaceX. They use Starlink in the “sea baby” USVs that attacked Crimea just last night.

https://tech.yahoo.com/general/articles/spacex-spends-signif...


GPS uses DSSS, not FHSS. The military version has a higher processing gain than the civilian signal, but it can still be overpowered by sheer signal strength.

I'm not saying that that's trivial against moving targets in a large area (especially if they can use directional antennas), but it's still very possible. GNSS jamming is a big concern of the militaries of the world, and there's currently somewhat of a renaissance of high-precision inertial navigation systems as a result.

Jamming a stationary satellite terminal, if you can get reasonably close, doesn't seem harder than cutting a fiber (although as a sibling comment has mentioned, jammers have the nice property that communication is restored once they're disabled, unlike damaged cables, so maybe the two can complement each other?)


Terminal side downlink?

With some satellites, you just point a dish at the satellite and get the same data everyone else gets. With more advanced ones, you have to be in roughly the same place as the intended recipient because the satellite has different antennas pointed in different directions. In either case, it's presumably encrypted data so what good is intercepting it?


You call that progress? I gain next to nothing from cheap reliable fiber being available at your house either.

Who would guarantee that the account has been used by the same person all those 18 years?

The same one who guarantees that it's really your ID: nobody.

A person using my ID needs to at least remotely look like me. A Google account is effectively a bearer token for anyone that can authenticate to Google, which is something that can easily be transferred.

There are significantly fewer concerns about symmetric encryption, and while it doesn't scale to the size or budget of a service like Signal, it's exactly the type of thing the military is good at:

Distribute a bunch of physical artifacts (smartcards) across the globe; guard a central facility (a symmetric key exchange center) extremely well etc.

The military can also afford to run its (encrypted or plaintext) communications over infrastructure it fully controls. The same isn't true for a service provided out of public clouds, on the public Internet.


The difference is in both culture and topology.

TLS (or more accurately, the set of browser-trusted X.509 root CAs) is extremely hierarchical and all-or-nothing.

The PGP web of trust is non-hierarchical and decentralized (from an organizational point of view). That unfortunately makes it both more complex and less predictable, which I suppose is why it “lost” (not that it’s actually gone, but I personally have about one or maybe two trusted, non-expired keys left in my keyring).


> They will EXPLORE collaborating on ACCELERATING development.

Compared to "FSD this year", every year for the past five years, I honestly find the approach pretty refreshing.


5 years...? I bought mine with FSD in 2018, and that was years after it was "right around the corner." Worst Kickstarter of all time... Though I do like the car itself.

Yeah, FSD was promised with the release of the first model S, 2012 - to be ready by "next year"

As such, it's been roughly 12 years just around the corner for Tesla and Musk enthusiasts.

To be more specific: Musk explicitly said in that marketing event that buying anything other the Tesla wouldn't make economic sense, as they'll earn their own price back as self driving Taxis within the following two years.

Still blows my mind that people believed him anything as these kinds of unrealistic promises were at the heart of every event since the start.

Believing someone with promises like that is - from my perspective - begging to get scammed.

From that point of view, people should be glad the delivered cars were decent. Most purchases with outlandish promises end with merchandise that is borderline unusable.


> Compared to "FSD this year", every year for the past five years, I honestly find the approach pretty refreshing.

As an 11 years Toyota driver I agree.


This is at the other extreme end though. They could do nothing and call the agreement to explore satisfied. Would rather they wait till they've removed at least three of the hedging words.

this is just the opposite extreme of "FSD this year"

Doesn't Waymo already have FSD?

FSD is something you can BUY and USE right now. Toyota is writing an MoU. They’re not the same.

Waymo is something that actually drives itself and you can hail right now. FSD is not actually “full self-driving”.

Tesla doesn't even call it that, there's “(Supervised)” after that. That's like calling it “Full Self-Driving (Not!)”.

FSD can’t self drive on any street though. Its just L2 like cruise control and lane centering.

I use it every day on every street in Vancouver. Where do you come up with your information?

If you car crashed while using "FSD", who would be liable?

You’re driving. Why you can’t watch a movie like you can on a Mercedes L3 car.

The ODD for drive pilot is so limited, I don’t think it’s really comparable. I have very little faith that their approach will scale to anything more than a traffic jam pilot gimmick.

It’s fair to argue that FSD is limited as well but I believe their approach is much more scalable.


FSD is not approved to work on any street. Drive Pilot is approved for highway speeds in EU. Limitations is what the government allows.

It's not just speed, its things like requiring lead vehicles, only highways speeds in the far right lane (likely for localization), weather, etc. The approach drive pilot is taking, with their technical investment and sensor suite, will not scale to urban/suburban driving.

FSD, in the states, works on approximately every street, parking lot/garage, etc.


What you're buying is not driving, by itself, fully.

Mine drove me from my house to the airport without my ever touching the steering wheel so what exactly do you mean?

If you can't sit in the backseat and watch a movie, it's not self driving.

All I’m saying is that starting FSD from park in my driveway and having it drive to my destination with my hands on my legs and then having it park itself when it gets there seems reasonable to call “full self driving” to me. I pay for the subscription and I would continue paying if it never got any better. I do live in a rural state, so maybe that’s why it works so well.

Having used cars that had that "supervised" driving feature.... Gosh, I hope you were paying 100% attention the whole time of that driving experience you described. Even the smart cruise control features I've used allowed my mind to drift, and I was glad for the beeping from the steering wheel telling me to pay attention. I don't use those features anymore.

If it's full self driving, then I assume that Tesla is paying for your insurance and taking all responsibility for any crashes it causes in your car?


Let's see it do that in the snow, heavy rain, anything that doesn't replicate ideal conditions in SoCal. You're riding on the sweet spot of a Gaussian and at some point you're going to experience an outlier when the machine makes a wrong interpretation of its inputs.

It sounds like you're recklessly interpreting the parameters by which your "self driving" car was made available to you

Did it also drive itself back to your home empty?


Adaptive cruise control could do that too.

How impressive it is depends on where you live.


> Adaptive cruise control could do that too.

No, it could not.


Do you even drive? Or have you tried using any of the features discussed above. I think I’m going insane seeing people comparing cruise control (lol) to FSD. One is a line follower, the other is a teenage driver with a fresh license. They’re not the same.

Oh please, just because it can do (bad, sometimes horribly dangerous) turns doesn't mean that 99.9% of the time it's not just a glorified cruise control with lane centering.

I owned a Model 3 with FSD. I own a Mach-E with BlueCruise. They're equivalent for the majority of drives I've done.


Except there’s FSD videos everywhere on X with every minor release, demonstrating the progress.

They edit their videos to remove the mistakes. It's all a lie if it only works 90% of the time and you don't know when it's going to fail after being lulled into inattention.

Waymo made a very good point about this, fundamentally that's the problem with a progressive rollout - drivers will stop paying attention when the software is "good enough", and the result will probably be crashes. So for them, it was all or nothing - either it fully drives itself, or it's not worth deploying.

They constantly highlight regressions.

Yes, that’s called marketing. Believe it when they do what Mercedes does and accept liability rather than trying to shirk it.

If marketing is a 20 min video demonstrating both progress and regressions for each minor rev, then I’ll take more of that.

I'll believe it when they accept liability for the actions of their autonomous vehicles. Videos mean nothing.

You fell for Mercedes’ marketing so that’s hilariously ironic

I’m sure that sounded more clever in your head but an actual legal agreement is more than marketing.

Mercedes system is dumber in every way, they merely set the criteria narrow enough to get dibs on L3 for gullible people to repeat on the internet. It’s OBVIOUSLY more dangerous than FSD under the same circumstances.

https://www.reddit.com/r/SelfDrivingCars/comments/1cpwir3/te...

So which company of the two is the one endangering people


No, they really are no different. A legal guarantee doesn't actually mean the car is safe, it means they will pay for it when the safety features fail. Those fees paid out can just be considered a marketing expense to make the car appear safer.

it means they will pay for it when the safety features fail

Tesla will not - it will blame the human driver for not paying attention. That's the difference.


This is misinterpreting what I'm saying. I'm not arguing for the safety of Tesla's system. I'm saying that judging the safety based off of corporate marketing decisions is a mistake and putting a guarantee on a product is a marketing decision.

Even if this originated as a marketing thought bubble, there's no way that such a decision could've been made without direct approval from the executive (including the CEO), and only after taking advice from their general counsel and consulting with the board. The potential reputational damage is too immense for such a decision to be made by "marketing" alone. What you're describing has happened before and the courts awarded massive punitive damages against the motor company.

https://en.wikipedia.org/wiki/Grimshaw_v._Ford_Motor_Co.


Yes, they are different. The degree depends on the company and if they have a history of trying to weasel out, but a legal agreement makes it harder to dodge liability in court.

Tesla would love to offer the reassurance to buyers but there’s a reason they haven’t done so: they’d lose money on it.


You are failing to understand what I'm saying. They don't have to weasel out of legal liability in court. They just bake the legal settlements they know they will have to pay into their marketing budget.

Has no one watched Fight Club and heard the anecdote about how a company will only recall a car if the cost of the recall is lower than the cost of settling all the lawsuits? All this guarantee tells us is that Mercedes did a similar calculation. Taking legal liability is not proof the car is safe. It is proof that they think the value of customers thinking the car is safe is more valuable than the cost of paying out settlements. Tesla not making the guarantee does not prove their cars are unsafe. It is evidence that if they did the same calculation, that got a different result. Maybe that is because the car is more dangerous, but it could also just be a different marketing philosophy and Tesla notably does not approach marketing like most other car companies.

The conclusion that you reached in which the Mercedes is safer than the Tesla is valuable to Mercedes and that opinion was indirectly purchased by Mercedes paying out legal settlements.


Yes, we know that companies exist to make money. My point is simply that when they are willing to make a stronger legal commitment in a country famous for litigation it suggests that they have a higher confidence level in their system.

Think about it like this: company A says “our government product is military-grade. We have a 1 year warranty.” while company B also says their product is tough but offers a 5 year warranty. Which one do you think has better data supporting the durability of their product?


That scene is a reference to Grimshaw vs Ford Motor Co.

The precedents set in that case mean that the liabilities arising out of legal action based on 'strict liability' are likely to be extremely punitive (these days, well upwards of the $147M awarded against Ford in 1980, and into the billions). Any company that did not factor such a payment in their calculation in addition to the indirect costs of reputational damage, deserves everything they get. I doubt this is the case with Mercedes.

One question I do have that perhaps someone here will know - is the Mercedes guarantee limited to certain locales? e.g. Germany only as the roads there are in good condition and well marked? (I'm assuming here).


Yes, and them putting their money where their mouth is means they are reasonably certain the accidents wont eclipse the profits, which is a much better signal than whatever Tesla marketing is putting out, seeing as they're not able/willing to do the same.

Money talks, as it were.


I'm not familiar with the facts of the matter but if it is indeed the case that Mercedes is indemnifying drivers for accidents caused by FSD, then that's far more than marketing, and your comment (without presenting any facts to the contrary) is unwarranted.

The bigger difference is that Mercedes’ system only works on highways, under 40 mph, and you need a car in front of you that it essentially follows.

It’s for traffic jams, and only usable in them. There’s not a big legal liability when the biggest risk is probably a fender bender.


> The bigger difference is that Mercedes’ system only works on highways, under 40 mph, and you need a car in front of you that it essentially follows.

And geofenced to specific highways, only during the day and during good weather.

It's still cool (to me at least). But it's bizarre seeing people dismiss FSD as being the same as adaptive cruise control while touting Mercede's Drive Pilot. Drive Pilot is a lot closer to adaptive cruise control than FSD.

It's unfortunate that there's so much misinformation that gets thrown around whenever this topic comes up.


People are interested because Drive Pilot is L3 while FSD is L2. People are naturally more interested in the more advanced systems, but that would include FSD if Tesla can improve it to perform closer to the way it’s been marketed. Exaggerating the capabilities for over a decade juiced their share price but it also gave them a reputation for failing to deliver which is going to need hard data to shake: putting their money on the line would be one way to do that.

Progress would be get certified for self driving. For comparison, Mercedes, BMW, Honda etc have L3 cars on the market. Mercedes just got approved full highway speeds in EU and working on L4 certification.

I just checked out Mercedes, and it appears to be geofenced with a lot of restrictions[1]:

> DRIVE PILOT can be activated in heavy traffic jams at a speed of 40 MPH or less on a pre-defined freeway network approved by Mercedes-Benz. DRIVE PILOT operates in daytime lighting conditions when inclement weather is not present and in areas where there is not a construction zone. Please refer to the Operator’s Manual for a full list of conditions required for DRIVE PILOT.

Only on select freeways and only under 40 mph (and only during daytime with good weather conditions) sounds like it wouldn't be particularly useful.

Still, the tech is cool, and moving in the right direction. It's just always hard to really tell the state of things without doing some digging, because there's a ton of misinformation that gets thrown about whenever this topic comes up.

[1] https://www.mbusa.com/en/owners/manuals/drive-pilot#2


That’s what the government allows them to do. They got approved for freeway speeds in EU and are trying to get approved in CA. You need to prove that the car is safer than a human driver.

There's now a thing called "FSD", yes. But it's not FSD as in Full Self-Driving, as in L4. It's still an L3, the driver still needs to be at the wheel and paying attention. "Full Self Driving" implies L4. What Waymo has, with no one at the wheel, is L4.

FSD is Level 2. The driver must, at all times, be supervising the car. L3 implies that the car, in exceptionally ideal and limited circumstances, takes over full control unsupervised.

Even if you were to take Mobileye's definitions, it would still be equivalent to other companies at best (hands off, eyes on).


If progress is noticable in a 20 minute video, Tesla has a long way to go.

As do Waymo and Toyota

Who is blaming hackers? Not the government or network operator at least.

In fact, you are doing exactly the thing they have avoided so far: Jumping to premature conclusions.


I haven't wrote that the government said that. But a lot of online commentators and even some news outlets immediately started to discuss "cyberattacks" (including the linked article) and asking "maybe it was the Russians?".

Most people who are familiar with how electric grids operate are in agreement that high degree of renewable generation has likely made the situation worse not better. We know that right before the blackout the grid had a sudden dip in frequency, which is exactly the condition which leads to unstable work of inverters used by renewable plants. Just watch this Practical Engineering video [1], if you are not familiar with the topic. It discusses exactly the problems I wrote about. So renewables may have not been the direct reason for the blackout, but they are highly likely have helped to tip the gird behind the critical point.

[1]: https://www.youtube.com/watch?v=7G4ipM2qjfw


A lot of news outlets were not part of the left leaning side of the goverment; the theory for an attack was set from every media, from the Basque/Catalan left and right to the pro-Spain left and right.

> stores would pull out the credit card imprint machines, but those don't work anymore because cards are flat

It's the other way around: Cards are flat because a carbon imprint doesn't afford the merchant any payment guarantee by the card issuer anymore anyway. (In other words, the "floor limit" above which cards require electronic authorization is now zero.)


At that time merchants probably still checked the signature on the back of your card against your signature on a bill. These days nobody bat an eye when I use my unsigned card. I guess at that time a matching signature would give the merchant enough confidence to process the card offline.

The merchant's confidence is irrelevant if it's not backed by a guarantee of the scheme (effectively forcing the issuer) to pay even in case of fraud.

The people operating these imprinters are sales clerks and waitstaff, not graphologists or experts in detecting altered physical credit cards. The sophistication of fraudsters has also advanced, and as a result, a system that might have been good enough in a pinch 20+ years ago isn't necessarily good enough today.

That said, in my view there's no excuse to not leverage the physical chip present on effectively all credit and debit cards these days, which is technically capable of making limited autonomous spending decisions even with both the issuer and terminal offline in scenarios like this. It probably won't happen without regulatory pressure, though.


IIRC tapping credit cards on London’s underground turnstiles use offline chip authorization to avoid delays.

It uses offline authentication (i.e. checking whether the card is authentic), but not authorization (i.e. whether it’s funded, not stolen etc.)

Unfortunately, too many cards, and all mobile wallets, don’t support offline authorization for that to be viable.


I started signing my cards with an all caps "ASK FOR PHOTO ID" 30 years ago. It raised a few questions when I would travel to the US and use them there, but was never refused a transaction.

That’s an urban legend, and stores are not required to actually do so. (And as far as I know, a thief could sign the receipt in all caps, ASK FOR PHOTO ID, and it would be a valid signature :)

In fact, even verifying the signature is no longer required in at least the US.

Signature verification also only solves cardholder authentication, not card authorization (i.e. figuring out if the card is funded, still valid etc.)


I've been signing "CHECK ID :)" for several years now and it's only a few times a year that someone notices.

Join us for AI Startup School this June 16-17 in San Francisco!

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

Search: