Scaling Blockchains - Computerphile
Вставка
- Опубліковано 11 вер 2024
- Scaling up blockchains is far from simple - Dr Shehar Bano of the Information Security Group at UCL dicusses some of the problems.
UCL Link: bit.ly/C_UCL-pe...
/ computerphile
/ computer_phile
This video was filmed and edited by Sean Riley.
Computer Science at the University of Nottingham: bit.ly/nottsco...
Computerphile is a sister project to Brady Haran's Numberphile. More at www.bradyharan.com
The scaling part starts at 13:13
I admire that Dr Shehar Bano manages to talk about something substantial without non-functional pauses, hesitations, or saying "uhm".
grieske more difficult than it seems
She gives a good interview. Well thought out. Interview her again!
Less like an interview, more like lecture IMO.
...Did you happen to read her Summary?
“My research interests include blockchains and Internet censorship. Broadly, I am interested in studying networked systems, particularly in the context of security and measurement.”
Dr. Bano gave the clearest and most succinct blockchain description and summary talk ever! Does she have other presentations you can recommend?
Perhaps a sequence of videos? I would very much like to see her start her own UA-cam channel, much like her fellow Computerphile alumnus Rob Miles.
I could listen to her speak all day. Clear, conscise, informative.
Just found out my .28¢ worth of bitcoin I got a while ago has turned into $8 worth. Not bad.
This was a very good and realistic explanation. It's the first video I came across which has described things the way they are.
I was hearing a lot about blockchain, and all were portraying it like it's the perfect thing. And the only thing bothering me was, how can they deal with latency so perfectly? I mean that's one of the major challenges when dealing with shared data. And if there was no problem with latency with current technology, then existing systems can also work equally well.
Secondly, what about scaling? Where would all the data go? In the age of cloud, where there's a probability that even OS would be in the cloud and we'll be just holding devices, there's a concept that all the world's data would be in our systems? Not possible of course. And if only some peers will handle it, then we're again back to where we're coming from.
But this video explains the goods of blockchain, and how there are some challenges which they're trying to deal with.
Nice! One of the best, well laid out explanations on blockchain that I have heard so far... for my level of understanding I mean.
Dr. Bano is smart, tells the story really clearly, and I could listen to her accent all day. But I smiled when she’d say we don’t want to get forked. No we do not!
She's very eloquent. Great interview.
To implement an electronic voting system on a blockchain infrastructure u need to address the issues 1.)the validation process of the voters and 2.)by doing that keeping the anonymity of the voters as well.
Very fine! She seems well prepared and watches her pronounciation very carefully. I understand her by far better than some scotsman talking way too fast :-) One of the best videos lately. And actually a lot of content again (for a change ;-/ )
I've got to say, this topic is massively interesting and the video was awesome. I was totally glued to her every word. Thank you numberphile!
oops, that should have been computerphile...
Bitcoin's blocksize is no longer capped at 1 MB, with SegWit update activated in Auguest the maximum possible blocksize is 4 MB, but on practice it would be close to 2 MB. We've already seen a 1.3 MB block.
Yeah, let's give these "leaders" such a big stake in the system that they won't want to misbehave.
That worked *so* well with banks.
What could possibly go wrong?
Great interview with a very intelligent woman! I could watch her talk about Blockchain all day! 😍
I strongly agree. The manner of her speech, whether it is an accent or just a focus on articulation, it is quite pleasant to my ears
Yeah, block chains scale up linearly. More transactions require bigger blocks, this is why all the focus is developing off chain settlement so that we can take 99.9% of transactions off the blockchain.
From what I remember, blockchain transparency is the issue with voting, as it makes voting traceable, eliminating the secret ballot. Of course, many attempts are being made to eliminate this problem win various ways, but this can add additional problems.
For example, one proposal I just read can be taken over by only a few people organizing the vote, thus relying on them to be honest.
This video has more than tripled its value over the last 15 hours.
Wow stunning woman and very clear explanation of something quite complex. I've never really understood how cryptocurrencies work but this cleared up a lot.
do an episode on zk-Snarks next!
Very good explanation. 10/10 ! Thank you Dr. Bano!
Sheharji the way you have explained, I got reminded of my friend who used to teach me in college...plain, simple and in easy language
Lady has done her homework, thanks for all the info.
Love that "Practical Visual C++ 6" book on the shelf.
I see an other huge issue with block-chains' scaling in numbers as people want to use more of them for various uses. The integrity of block-chains relies on the race between "miners" but if any one "miner" has overwhelming computing power compared to the others, it compromises it.
For example, let's say there are a 1000 different block-chains. That means at least half of the block-chains have less than 0.2% of the worlds mining power devoted to them. So if one miner with 1% of the worlds mining power were to focus on a weak chain, they would be able to compromise it.
That means at any given time, there is a limited number of reliable block-chains, depending on how big a share of the mining power the biggest mining entity has.
If the biggest mining entity has about 1% of the world's mining power, that means you can probably not have more than 50 different reliable block-chains in the world. And since the throughput of a block-chain is limited as explained in this video, that means the overall uses of block-chains are very limited.
Proof of Stake 'mining' can get around many of the problems of Proof of Work cryptocurrencies like Bitcoin if implemented intelligently, including the problem of centralised computing power.
Have a search for the 'Ouroboros Blockchain Protocol' being developed for Cardano by IOHK.
Edit: I should really watch the whole video before commenting :) Still, check out the Cardano project and Ouroboros.
This is the PoW dilemma
This assumes that a big miner can use their hardware for multiple hashing algorithms, which is often not the case. This is why it is so hard to strike that balance between ASIC centralization and botnet vulnerable ASIC-resistance. I totally agree with the idea of there only being one viable, say, sha-256D coin, though. Still, maybe some viable form of Stake or Storage backed algorithm will come along.
If the question is blockchain scalability, why not discuss storage constraints? This is especially an issue with something that's transacted as often as a currency.
It's not really an issue, space is very very cheap, in theory you only need a few archive nodes to keep the whole chain, most nodes only use a few blocks at a time. 1mb is about 2000 transactions. Visa handles about 2000 transactions per seconds so 1mb per second added to the chain, that's about 3 petabytes for 100 years of visa-level transactions.
Transactions don't include block ids (hashes) when referring to previous outputs, they include transaction ids.
Great video. Really cleared my concepts.
Cleared your mind*, perhaps? Greetings from the USA. :)
This video is simply amazing about many of the issues which needs to be addressed due to lack of scalability of bitcoin desgin and blockchain technologies. As it passed 5 years, it could be good an update video of what has been solved or changed since then.
Well presented in an easily understood manner. Very informative. Thank you!
Very Informative
Seriously thank you for putting up videos like this its really helpful , and the Dr explains things really well .
Couldn't those "leader" nodes just be an alternating committee, and not a permanent group? That was collusion could not really be possible.
Simply the best video on blockchain description ever!
So the concepts are amazing, and blockchains are the first, very naive implementation of said concepts.
It is essentially nothing but a linked list with a protocol layered on top to ensure tamper-proofness.
Can't wait for a better implementation! Smart contracts have the potential to be completely transformative to our society!
When it comes to scaling up blockchains, solving some problems requires a change in the architecture. A tangle (or IOTA) might be an option for that. It still provides the basic features of a blockchain, but doesn't requires mining, hence enables micropayments in IoT scenarios.
Great video, but I don't understand the comparison between Bitcoin and Visa. Bitcoin is a currency, not a bank, you can trade it directly but normally you would do it through some third-party service just like Visa or Mastercard does for the USD and other traditional currencies, this solves the latency problem.
The subject of the video is fascinating, but can we take a moment to compliment the presenter? She has such a clear way of speaking. No "uhh"s, "well"s and such. I was like she had it written beforehand and was just reading. I was listening to the video while doing the dishes, and it felt almost like an audiobook. Amazing stuff, I'd like to be this good at public speaking one day.
This is one of the best explanations I've seen given for Bitcoin for a general audience. For once, it seems quite accurate, compared to what I know of the system.
I think this woman seemed a bit nervous at first, but got into it quickly! Very nice good video!
I would love to hear what she has to say, but I am absolutely horrible with accents - I can't even understand accents in my native language.
Could we get subtitles on this video? I know my request may be hurtful, but shit, if I could fix my sound processing, I 100% would.
Why not have a disposable committee? I think the instinct to trust the leaders is the problem. Instead once a result from a leader is used, blacklist that node for as long as possible. If I am being nefarious I have to wait a very long time to come become eligible again, thereby demotivating the the nefarious behavior.
There are now many competing technologies such as blockDAGs, Tangle and Hashgraph. All of which are Directed Acyclic Graphs. Tangle gets faster with scale and is only limited by bandwidth. Hashgraph has Asynchronous Byzantine level security and throughput over 250000 pre-sharding.
The best blockchain explanation I've seen.
Great interview! I'd love to see a follow-up video.
Please elaborate on how the IOTA Tangle fits in the mix. Does it solve the scalability problem? What are the new challenges we will face when using Tangle?
I like to hear her talk, shes intelligent and very nice to listen to. Thanks for all the info!
How do you index the blockchain so that you can quickly find all transactions involving Alice? Can this be done without every node having to store the entire blockchain?
You don't have to go back through the whole chain, only far enough to ensure that Alice has the money. In theory only a few archive nodes need to keep the whole thing. Also the whole thing is actually very small, bitcoin reached 100gb last year.
The video is very informative and she is lovely. Perfect mix
Finally someone actually talks about why blockchain is fundamentally bottlenecked. I have to explain this to everyone all the time. I personally don't hold out much hope for figuring out how to scale such a thing.
So here you discuss possible improvements to the blockchain, but can these changes be implemented on existing blockchains? Or everytime you want to change the blockchain a new genesis block must be created?
If you have to wait for a few blocks to be finished after you submit yours... just to know that it is final, doesn't that imply that your submission could be lost? How would this happen and isn't that a big problem?
I second that.
Well i think that if its lost, there is no transfer of ownership and so you'd just have to do it again? (I Might be mistaken or misunderstand the question)
If your transaction is written to a fork that is subsequently (within minutes) abandoned, then it again becomes an outstanding transaction. This possibility has been accounted for, so your transaction isn't lost ; it just got "rolled back" and needs to be retried. It _will_ go through, it's just a matter of time (and then waiting even longer to be sure it's permanently committed).
The transaction gets included at a later time, in a different block.
By about 16 minutes in, the concept I'm seeing is, that in order to make the Blockchain function better, we are inventing politics dare I say a political system for governing. It sounds like checks and balances almost. Have a group of nodes handle error checking, but then how are we sure those nodes don't cheat the system? Perhaps another group of nodes to check them? The parallels between blockchain and our own political systems almost makes me think that researching one would lead to understanding the other, bu maybe I'm stretching this too far?
George Erfesoglou "Checks and balances" has been a concept independently of government for a long time #whowatchesthewatchmen
Sounds to me like she's explaining the two-tiered governance system within Dash.
Of course with a blockchain currency the creator doesn't have to siphon the funds; they can just give themselves as much as they want in the genesis block. Of course everyone who joins will be able to see this so it's going to effect who decides to join.
I don't think you stretch this too far, that is what decentralization is all about, especially if we are talking about decentralization of power, which bitcoin is certainly a part of, since nothing says POWER like the creation of money.
I think she's potentially talking about DPoS.
fascinating and clear explanation. Keep going on this topic! would love to hear about the other decentralized consensus methods mentioned midway through.
If I pass some bitcoins to you, then why is some leader involved? I need to sign that they are now yours. Then get them added (is it this last bit, to stop people adding random stuff?). With forks, could they use merging? Could be a problem with double spending. Do coins have an ID or is there just a ledger of spends resulting in a balance?
They’re not, strictly-speaking, blockchains, but I wish she had discussed Directed Acyclic Graphs, which are a type of distributed ledger that do reverse the scalability problem.
We need scaling on second layer, sidechains => Lightning Network. Blockchain will become a settlement layer.
awesome video on blockchain scalability. very thorough and informative! computerphile ftw 😎
Proof of Stake blockchains scale much better than Proof of Work blockchains, because in Proof of Stake every wallet can create blocks, you don't need hashing power.
But still people keep thinking that you need to waste huge amounts of energy to "mine" blocks.
The committee approach has been tried with masternodes but that's also not a perfect solution because it decreases the decentralisation in the network, creating an elite of nodes that decide more than other nodes.
I think I finally understand it at last! :)
There is some sort of a very high-frequency noise appearing in the sound of the video at about 4:40-4:50. Am I crazy?
8mb blocks are totally acceptable by today's tech. My 2015 laptop can verify a 8mb block within 4 seconds.
If your hard drive can't hold that much data, you can just prune and leave like 5GB of blockchain on your hard drive. No body can rollback 6 blocks, let alone 5GB of blocks.
If you ignore the issue of latency.
Enter: Bitcoin Cash
Hi Im loving the video , but that pencil sound with the paper is killing me ... aggghh , i have to say i love the issue, the explanation is really clear and also love your channel. I have some questions ... if we want to have a server (database) using blockchain tech , i know it wont be a classic centralized server , but my question is ... as we put more info into the data base , the chain gets longer and harder to solve? is there a limit? ... is the limit the capacity by the minner's nodes to solve the hashes ? if the calculating capacity of the nodes get increased it makes the system could keep growing? also love to know if this system is moore secure than RSA codification, because if it is, and also is open source ... will be nasa getting really mad?
I heard that the Bitcoin Core team don't want to do the hard fork of 2MB block increase and want to implement the lightning network instead. How does the lightning network would actually help scale if it is a system built on top? you would still have 1 MB every 10 minutes. I agree that increasing the block to 2MB is just an immediate solution and not a long term one because in a few years we will be back to the beginning. Can you make a video on this lightning network proposal?
Surprised there was no mention of Lightning Network(s) and how they're a very compelling approach to scaling blockchains...
lightning is directly dependant on the speed at which you can add blocks(you need to add to the blockchain to create the channel). You can't open/close channels fast enough. You can maybe accommodate a few million people on the lightning network at current speeds so you are off by about 3 orders of magnitude to handle the entire world. Just opening a single channel for every human at current speed would take forever considering how many humans are born every day. So you are back to square one, which is how do we handle more transactions on the blockchain to allow for more channels opening/closing.
Lightning Network is not scaling Bitcoin, it is a separate network trying to solve a problem that an on-chain scaled Bitcoin mostly provides. It has its use for micro-transactions which are not practical on-chain no matter the block size, but it does not scale the blockchain, it only settles to it. This video is describing scaling the blockchain itself, which by the way is still necessary in the event Lightning or similar networks garner wide use.
The annotations at the end of the video, which link to the Krack attack and botnets, are missing.
Great interview btw ^^
I wonder how the idea of a hierarchy of sharded committees works or not.
Like, if the committee that handles A-F needs access to G it could pass it on to a higher committee which handles A-N, with the highest committee handling A-Z.
pretty much confirmed my suspicion that the #1 problem is scalability. 7 transactions per second is indeed very very slow.
Brady could you please do a video about Monero? The Monero blockchain is encrypted and that's pretty damn cool.
Very nice explaination... for convoluted concepts of BlockChain technology
The blockchain is currently 65 GB. I don't have the disk space to use my wallet?
this video just answered all my questions about the block chain thank you
Very clearly and beautifully explained. Thanks!
great video, thanks to this promising scientist
Great!
Have been waiting for this episode
13:05 what study is she referring to?
More Dr Bano!
What's your take on PoC (Proof of Cooperation), used by Faircoin?
Fantastic discussion on the topic. Thanks!
I don't see why using shards would be too different than purposely allowing the blockchain to fork, then implementing some method to merge those forks back together.
Don't invest bitconnect...
MisterDream Fx yup, it's extremely dodgy that they have a hidden trading bot to fill in the gaps.
How blockchain pruning impacts on the scalability concerns
Seem like Blockchain is totally in a bubble right now... Their value is rated soo high, but the reality is that this technology is yet to be able replicate the conventional banking transaction rate, and there's more work to be done.... Of course there's plenty of room to improve, but lets hope Blockchain users are wise enough to spend their 'bubbling' money on R&D of Blockhain infrastructure rather than just sitting on it... the money is like an IPO, you have to quickly utilise it to create a product rather than just letting it 'rot' when investor realise their investment is just hype.
Awesome video !!
Some blockchains have a 15 second block time instead of 10 minutes. How does that work?
Some, just because they're not very popular. Others, because they're fundamentally more advanced (say, Ethereum)
They just have more sophisticated protocols to deal with stale blocks(example: bitcoins keeps the longest chain), the shorter the time the more stale blocks you will get and the more sophisticated an algorithm you will need to deal with them. See GHOST protocol (ethereum) for an example.
Excellent explanation!!
Solid explanation. Thank you.
Very good explanation , thanks!
Clear, deep and accurate!
This was good but I miss the brown paper.
How would the blockchain or bitcoin work in interstellar travel and thusly latency at the speed of light, or if aliens with supercomputers visit and solve entire blockchains at the expense of us puny humans?
The more I learn about bitcoin and blockchain the less I think this will ever be able to scale to the requirements of a widely used currency.
Very nice video!
So why not just form a super-committee out of the committees governing the segments / blockchains for the time it takes to resolve a transaction that needs data from those segments / blockchains? Anyways, that's my 0.0002 satoshi ...
It's not Charlie, it's Bob!
Thats was the topic of my thesis! :D
Fascinating
Interesting.
Thumbs up for Visual C++ 6
The price of bitcoin went up by almost a thousand dollars since they filmed this.