Quote:
Originally Posted by Kazuya
Ok so this guy is a toxic clown. Let's deconstruct it.
I'll address the last 2 paragraphs cause he's specifically talking about Eth.
Eth can handle much more than 15 T/S. Did he pull this number out of his ass? And second off, comparing a BTC transaction to an ETH transaction is like apples and oranges. BTC sends just bits of little info from A -> B. That's it and that's all it can pretty much ever do. Eth can do that, but it can also send tokens, and it can execute logic on those tokens or information that it sends. Transactions like that do take more gas. Those ones [transactions] that BTC do, it would take probably 1 GWei (10^9) - which is roughly probably ~2c to complete (and probably 20-30 seconds at ~215k daily transactions which is what it was done at today).
Miners can also decrease/increase the gas limit, so if **** is hitting the fan (say like on the day the BAT ICO happened) they increased it and yeah it cost more, but the logjam went through fine in the end. They can also decrease it, which Vitalik asked them to do a few weeks ago (45% dropped acceptance to around 1 GWei from 20 when it was beginning to get busy). Why would they do that? Cause they're not idiots and realize it's good for the ecosystem. Scaling issues will be worked on and I'm looking forward to future forks, but for him to say 200 gamblers will somehow jam Ethereum network is complete fabrication and will NOT happen (nor has it ever happened).
For the sake of trying to be more objective I should have put an asterisk on this sentence and explained this in much more depth.Technically, with where Ethereum is this moment, this isn't yet changed and I shouldn't be claiming otherwise as if it's a 100% lock to happen. It's going to change in future updates in the way I describe below, but as of this moment in time it's not correct and ~15 T/S is the limit*. I'll do my best to explain what I mean about the whole T/S debate but I want to let everyone know I am not a developer. I'm writing this to try with the purpose to try and help non-technical people understand scaling/sharding with Eth. PoS will have an effect too (using more computing power towards running smart contracts & securing network instead of hash #'s, but best left for a different post) If someone is a developer and has the technical prowess to put this in easier-to-understand laymen terms (or sees any mistakes I wrote here), obviously please feel free to weigh in and contribute if you're willing!
Anyone, I will try my best to better explain here:
Imagine you have a single blockchain and for simplicity sake there is say 100 T/S in the last block (happening currently at 15 T/S).
What causes congestion/scalability problems is well what happens when there is more T/S needing to be processed due to increased usage? For example, just in the last couple months Eth daily transactions have grown from 70k to 215k*. With BTC this is what is leading to higher fees/delayed logjams (more transactions than it can handle in a timely manner). So Ethereum, if it keeps growing without updating the platform, will have the same issues down the road. As of June 2017, the platform is fine for the # of transactions daily (and should be pretty good til maybe 700k ish without any significant delays) So right now, we can have more blockchains (with say 100 T/S) all right behind the first one, but eventually that's going to become a problem. (Say it's like imagining a train with a bunch of containers on 1 railroad track trying to pass through a finish line)
Under current scaling laws that's what would happen. Now Metropolis (the next update) it will likely address improvements that make it easier to interact with singular blockchains (like the one with 100 T/S described above). It will likely have Z-Snarks which add privacy layers through greater anonymity, and make it easier to build Dapps on. It will still be 100% PoW, and not do much, if anything, to address true scaling issues.
After Metropolis, and before Serenity (which will be 100%PoS), there will be an update that switches from full PoW to a hybrid version of PoW/PoS to almost serve as a 'test run' of sorts for what's to come. This is where scaling debate will come to the front and center because for the first time the scaling 'laws' will change. This update AFAIK may have a form of sharding within it.
The way I understand sharding is it's a blockchain on a blockchain (as if it was parallel to the track, not behind it on a long singular chain). Right now let's try to visualize what the current PoW version of Eth looks like: Let's say there is 1000 blockchains, and each one doing 15 T/S. You would then make 1000 blocks in a 15 second period. Each of those 1000 blocks are cryptographically hashed (basically this means using computing power to spit out an individual identity tag consisting of characters), and in turn each of those 1000 are then sent through that railroad track individually 1 at a time (or 15 per second). This isn't very elegant because it takes more time and power and can't sustainably address long-term scaling.
Sharding changes what this would look like: Let's say once again there is 1000 blockchains, and each one doing 15 T/S. You would then make 1000 blocks in a 15 second period. Sharding allows you to hash those 1000 blocks into say 10 blocks. It's taking those 1000 cryptographically hashed blocks, and condensing them into just 10 cryptographically hashed blocks (that if you were to dehash would contain the 1000) Thats ^10 effect on scaling. Visualize those Russian dolls but on steroids that would stack within each other. 1000 -> 10 -> 1 would be another level of sharding. So in other words, that 1 blockchain would have the information of 1000 blockchains contained within it, with it's unique hash. That 1 blockchain would count as 1 T/S, yet if you were to de-hash it, it would technically be 10,000 T/S. So yeah, in the example, imagine instead of a very long singular train track, we now are building out parallel sets of train tracks. All those tracks, containing all that information, can be condensed with a ^10 effect (for sake of this example) on scaling. We're no longer building train tracks linearly, but instead exponentially. When you hear ETH people talking about about it becoming the next super computer of sorts, that's what they're referring to. When usage goes up, all we'd have to do is add another ^10 layer of sharding. If this works as planned, pretty soon, a toaster talking to your iPhone is within the realm of possibility and if that were to happen it wouldn't cause any sort of congestion on the network. Hope this helps!
Cliffs - Paragraphs explaining why a toaster could theoretically talk to an iPhone in the future
*
https://github.com/ethereum/wiki/wiki/Sharding-FAQ
*
https://etherscan.io/chart/tx
Last edited by Kazuya; 06-11-2017 at 04:39 PM.