Hacker News Reader

Comment by helsinkiandrew | Hacker News Reader

> The data that each smart meter will transmit will be validated and certified by the blockchain, so as to be able to unequivocally guarantee the savings obtained at a certain point in time

Block chain doesn't validate things happening in the real world (energy saved, KWh consumed) only auditors/inspectors can do that. The data from the smart meter can always be faked before it goes to the blockchain network.

Have I missed something? I'd be inclined to give Woz the benefit of the doubt on most things but I don't get what blockchain gives here apart from marketing BS - for which its a year or two behind the times?

View on HN

You've hit the nail on the fundamental issue. Authenticity check at the Physical -> Digital touch point. Some efforts such as digital KYC for banking are able to solve this because the process involves a government issued identity, so the authenticity of the physical -> digital transmission is sort of embedded into the physical asset itself. Tampering the physical asset can be quickly reverified using several of the asset parameters (photo, DOB etc).

For aspects such as meters or constantly changing physical data, I am not sure if it will ever be possible to ensure the authenticity of physical -> digital touchpoint.

Is this something an embedded processor can help resolve ? In essence eliminating the possibility of fake data emerging from the meters ?

Indeed this is the crucial part.

Usually a company will not generate its own energy. Depending on the country, the infrastructure is sometimes publicly owned and managed.

Maybe the physical -> digital touchpoint here could be outside the boundaries of the company. The energy grid should report those numbers as references, to subscribers agreed upon by the company for example. There it could be certified by the government.

Or something like that?

Edit: Well of course now there would be an incentive to burn petrol to run turbines and generate locally, then turn them off to capitalize on a high in the energy efficiency security market...

Same problem with supply chains. There is still the need for a trusted party at the beginning of every block chain with a physical component to assure physical and digital data are the same.

* only auditors/inspectors can do that *

exactly. Everytime with this smart meter comes up, this is the crucial thing that just doesn't make sense. No replacement for "auditors/inspectors"!

This is called the oracle problem, and it's a very real problem and why I'm incredibly skeptical of "supply chain" blockchain applications.

This is why i had hopes for AI blockchains as the data and trained models or functional algorithms could be verifiably provable as increasing usability/productivity/efficiency etc. You could still say your device/robot didn't improve, but if you continued using the models or algorithms then that would be noticeable. You would have to know they were still being used though, so you would need a time-sensitive encryption that calls home to activate, and if used in multiple instances you'd know that too. But reverse engineering the encryption to get the secret sauce sounds like a workaround. It could be made difficult for average folk, but not impossible. The real advantage would be micro/nano-payments and fee-less transactions since the device is the "miner", ie. by not using BTC/ETH etc.

At the end of the day a subscription model is probably easier.

AI blockchains? Encryption that calls home? Wut?

https://github.com/steven2358/awesome-blockchain-ai

(works in progress, some more like app stores)

Is there a single case of a blockchain-enhanced service where the end result is provably better than a centralized system? I'm cynical given that every case I've seen so far has turned out to be rubbish, though admittedly I'm not heavily in that space.

Distributed energy grid because you don't want centralized taxation, and it's tangible.

The step change in software/OS from boxed to subscription will continue to next phase. With AI the current subscription model is probably not granular enough. Every device/robot will switch tasks and capabilities and upgrades/downgrades with such a frequency that traditional subscription breaks (refunds/tax/fees etc) so blockchain subscription per device without human interaction will probably work better. AI is bots learning from bots which is already distributed, so centralizing that may be inefficient (high bandwidth).

There's some defi apps that allow you to buy non-custodial versions of synthetic stocks 24/7. I would say this is provably better than the current centralized system. If all stocks were issued as tokens, then the velocity of money would drastically increase. In addition, the tokens themselves could be allocated to automated contracts for liquidity and to receive yield for lending/burrowing. Peer to peer finance.

I always thought the Oracle problem was a variation of the Regular Expressions problem:

Some people, when confronted with a problem, think "I know, I'll use Oracle." Now they have two problems.

https://blog.codinghorror.com/regular-expressions-now-you-ha...

Give the smart meter a private key protected in hardware, give the chain its public key, and you can now verify things.

> protected in hardware

That's the fallacy of DRM. There's no real way to keep data from a user.

So given a locked iPhone, you are able to extract its private key?

Exactly. Why do software engineers (myself included) so often confuse what is possible with what is practical?

We do that because our job in general is to ignore current practical limitations. The line between practical and impractical changes based on what we make; the job is all about creating solutions that either make a goal more practical, or improving the efficiency of a standing process.

This isn't just a problem with software engineers, but most engineering disciplines. It's impractical to transport logs hundreds of miles for processing, until we create a system that can do it efficiently. It's impractical to keep detailed records of billions of transactions, until we set up a database and reporting infrastructure that can handle that load.

No, but you are able to hook another device up to the battery so that the iPhone is wrong about its own power usage and phones home with incorrect values.

iPhones are a 15 year effort by a trillion-dollar market cap company to produce secure hardware that retails for $1000, and still regularly gets hit with root exploits.

This is your model for securing electric meters?

Split your incoming power to two lines and put the smart meter on one of them and it will register half the power. Route power from your diesel generator through the meter that's meant to be recording your solar generated power.

No amount of private keys and digital block chain security is going to provide physical security

Can’t you do that today though? ie. Circumvent your meter? It’s illegal though.

If the only way to cheat is illegal and can be done on the Blockchain anyway then why use a Blockchain rather than a database?

Apparently was enough to dupe Woz and get this to the front page of HN. Another blockchain success story.

Residential its harder but an industrial/commercial property could possibly have multiple power inlines, a diesel generator for emergency power, and perhaps solar panels on the roof and batteries. It may have agreement to use power within specific ranges during the day for different prices and supply power back to a power company or neighbouring businesses.

Under those conditions it's always possible to divert power to 'adjust' the perceived power usage.

My point was that block chain doesn't validate anything - a private key will prove that the data (good or bad) is from the specific user, block chain won't validate that its good.

In most jurisdictions in the USA renewable energy generators are registered and audited by a third party, (like mrets.org where I work). Anomalies with generation data can be compared directly with data from the grid operators. Utilities will also require these commercial operations to have a "commercial energy meter" which are very expensive.

When a meter company catches someone cheating do we want them to be able to rectify in their database or fork an entire blockchain?

The potential for fraud which may be tried in civil courts is not a slam-dunk argument for mutability.

If we're relying on "what is legal" for validating data why do we need a blockchain?

If something being illegal is enough to solve your problem, then why do you need blockchain?

oooh, it's illegal! but I thought the libertarian blockchain utopia wanted to get rid of any entities enforcing law because that's soooo socialist?

But in this case the blockchain implementation will provide irrefutable evidence to support any investigation. It's not some magical hand-wavy substitute for investigation, but recording this in a tamper-resistent manner can be useful (although blockchains are far from the only way to do this).

No it won't. All the records will show is that X amount of energy was produced at time T. There's no way of proving that these values were legitimate.

There's plenty of ways of proving that these values were legitimate, but they're nothing to do with the blockchain element, that's the investigation side. What the blockchain element does is say, "X was definitely recorded by Y at T". It's a tool in the box. It's not the box.

Blockchain doesn't provide irrefutable evidence of anything except that the record wasn't tampered with after the fact (and it's a very expensive and wasteful way to do so). Blockchain does nothing to verify the correctness of the data it holds, only its consistency with the history.

You would need control of 51% of the nodes to submit a block with arbitrary data and override consensus. You can’t just add any hash to a block and add it to the network. In order to be the miner to add the next block, you have to win a competition to find a correct hash that solves a difficult math problem. Since there is no way to start with a resulting hash and work backwards to figure out what piece of data gave that hash, the Bitcoin protocol uses this feature to create its difficult math problem. The math problem stipulates that the first miner to produce a hash with a certain amount of leading 0s will be the winner of that block and be able to add it to the network.

>You would need control of 51% of the nodes to submit a block with arbitrary data and override consensus

This is only true for corrupted miners. That is not an issue in this case because the company controls 100% of miners (and therefore, you are trusting them not to write bad data).

It is completely irrelevant to the question of corrupted clients, which no blockchain can guarantee against. All you have to do is attach some bypassing wires to the input and output of your "blockchain-enabled smart meter" to write completely fraudulent entries to the history that can never be fixed.

> Blockchain doesn't provide irrefutable evidence of anything except that the record wasn't tampered with after the fact

> (and it's a very expensive and wasteful way to do so)

No it isn't. You're confusing blockchain (the algorithm) with Proof-of-Work (a method for providing decentralized consensus).

> Blockchain does nothing to verify the correctness of the data it holds, only its consistency with the history.

Correct. Nor does any other data structure. Blockchain is literally just a data structure, nothing more.

Generator power is quite expensive, so that particular attack is probably not profitable.

lol it has been done in spain before: https://theecologist.org/2010/apr/16/spanish-nighttime-solar...

That's a significantly different attack, with different economics. It involved selling electricity instead of just buying less of it, and profiting from government subsidies. Clearly that can work if the subsidies are high enough.

> Route power from your diesel generator through the meter that's meant to be recording your solar generated power.

the original comment was about selling mislabeled energy.

Not sure what attack you have in mind.

> Have I missed something?

The first chapter of the whitepaper? [1] There's two sides entering a contractual agreement around energy consumption.

[1] Section 1.2 - https://efforce.io/WP_ENG_V1.pdf

Right, but that's the "traditional" E.S.Co system. It doesn't explain how the Benefitting company gives value to token holders, it just says they pay the E.S.Co.

Maybe it's badly worded. It could mean that the history is cryptographically resilient against tampering in the future. Ie once your meter submits a reading, it becomes part of the blockchain and cannot be tampered with. Is this really that useful though?

Yes I also think this is not the best way to use blockchain because sensor data can be faked one way or another. The best way to connect blockchain to the real world is by using human psychology and so I think the best use case is still cryptocurrency.

You can use crypto to build communities around a shared incentive, this in itself is an extremely powerful feature which doesn't get anywhere near enough attention considering that every idea these days relies on shared incentives in order to succeed.

Scientists in industrial research labs are often required to keep journals, and sometimes get them signed periodically from others, so that in case of legal issues (usually patents), there is a clear paper trail. This is not a foolproof method, and multiple people can conspire to change the journals at a later date. However, it is a better system than where people don't do this.

Don't think in binary. Think, whether this system provides better validation/certification than where auditors/inspectors come by at the end of the year to check records. Perhaps, it's harder to systematically cheat the system if you have to announce your data to the world daily, then if you wait till the 11th month, and cook your books so everything works out.

I doubt this removes the role of auditors/inspectors, but it might supplement them.

All this makes me think is that the 2020 crypto bullrun is starting. Prepare for more "celebrity" tokens, BTC swelling past ATH followed by a massive crash that takes out a majority of the current 550 tokens on CoinMarketCap, and countless other DeFi tokens on Uniswap

You are right in the strictest sense, but I think blockchain is still interesting here. With a blockchain, there is a tamper-proof public record of who submitted what, and that can make disputes through a good old fashioned legal process more streamlined later on.

As usual, this marketing page is quite vague, but I'm reading it as the cryptocurrency equivalent of existing carbon offset markets. Both require auditors to validate the offsets, but Efforce just handles the market side.

I'm waiting for the tidal wave of people who think that a "trusted enclave" will fix this problem and can never be exploited or cracked... [0] [1]

[0] https://www.theregister.com/2019/02/12/intel_sgx_hacked/?utm...

[1] https://blog.quarkslab.com/attacking-the-arms-trustzone.html

> a year or two behind the times

How exactly?

I agree with you that use of blockchain on its own is not fully secure, but that can be mitigated for the most part through the same sorts of proof-of-work and proof-of-stake mechanisms and entities that Etherium, etc. use. with adjustments.

This might be exactly the time for Woz and co. to get into this given that there are others handling those problems, right?

> I agree with you that use of blockchain on its own is not fully secure, but that can be mitigated for the most part through the same sorts of proof-of-work and proof-of-trust mechanisms and entities that Etherium, etc. use.

None of that does anything to prove that the energy consumption data you are submitting is true.

That’s what proof-of-work and proof-of-stake are about. I’m saying that the problem has been solved and would need to just be applied to this use case.

Nope: those only apply to elements on the blockchain, not elements in the real world. You'd need proof-of-not-work for a real energy efficiency business.

> That’s what proof-of-work and proof-of-stake are about

Nope. No amount of "proof of X" will help if you put false data on the blockchain in the first place. They will do a great job securing the false data though. Good luck with that.

Reply on news.ycombinator.com