Topic: The Ethereum Paradox - page 40. (Read 99889 times)

Vitalik and co. Will shard gas all over your nonsense FUD.

And then the partition that has accepted the transfer of gas from a partition that it did not validate now will have all its derivative transactions (and scripts) subject to reversal if later someone shows evidence that the balance was a lie.

You are writing nonsense (and you should know it which is why I am not being entirely polite ... please don't feed the damn trolls because you are not careful to think before typing).

I agree in general, but I wonder if there is more to this than it first seems. Yes it is true that merging two partitions into a block will need the block producer to validate both partitions to construct his new block, but that's not to say he necessarily has to keep validating both - if the partitions split apart again afterwards, for example that wouldn't be necessary.

You could formalise this such that an inter partition spend requires the source and destination partitions to merge for the transfer block only, thereby defining a partial order and only incurring an ephemeral increase in validation costs.

Let me ask a stupid question here. How a user can have gas in one partition and not to have gas in another ? Are you saying that a user had spent his gas but it didn't get validated by the system ?

And then how would a user who has his gas in partition 536 run a script that runs only in partition 1279?

Hint: he can't. Thus you would have entirely broken Ethereum with your suggestion.


I already explained that cross-partition spending is possible (i.e. won't destroy the Nash equilibrium) only if validation of all partitions is done by all validators, which thus defeats the scaling advantages of making partitions (shards).

You even agreed with this upthread wherein you stated that validators can't decide which block to mine on if they don't validate all transactions in the block.

Edit: in case this isn't clear, let me remind readers and you that I had explained upthread that when validator that is responsible for validating only his own partition has to accept a spend from a partition he did not validate, then he has no way to know if that balance from the other partition was a lie or not. Thus he can't accept it, because his reward may be compromised if later it is shown that he accepted an invalid balance from a partition he did not validate.

All the balances would have to be local to the partition and not shared - effectively like a fork of ETH for each partition.

edit: I'm considering the possibility of whether an inter partition transfer of ETH would be possible if the partitions merged once during the transfer and then split apart again post transfer block.

For starters explain what do you mean by saying that gas can't be partitioned ?

Not really needed, just watch that ETH-DEVs video upthread and try to get just a faint feeling for how close they are to a proper solution for Casper.

I ask you first where is the code example in Satoshi's white paper which explains a 51% attack? Hint: there isn't one because it doesn't have anything do with any specific implementation code, but rather is a mathematical structure.

Why don't you try to articulate what parts you understand and what parts you don't, then I will elaborate once I see where your blind spot(s) are. If you are unwilling to do so, then I will presume you are trolling (from a newbie account with 2 posts) as are the other panic-stricken trolls.

But yet this FUD troll 'crypto genius in his own head' will milk it for all it's worth with very long posts that no-one wants to read

Its pathetic really,  an ill (mentally and physically) and extremely jealous (old) man pouring incomprehensible and entirely premature FUD on software that no one knows the inner workings of because it isn't even released yet.


Apparently casper wont work because of some obscure mathematical principles which only this self proclaimed genius is able to understand and is incapable of explaining.

TPTB_need_war its really hard to follow your arguments without specific example. If you understand this deep enough give us specific code example which will illustrate what you are talking about. Without an example it seems very high level abstract esoteric discussion.

Even that wouldn't work because as you pointed out, the entity running the script pays the gas, and that entity can not in every case have its gas balance in the partition that validates the script.

But gas never was going to work economically any way, because all the validators need to be compensated, not just the one who wins the block. Any means of equal sharing the gas between all validators will be Sybil attacked (again any arguments about changing this reality via consensus-by-betting would return to the fact that proof-of-stake or deposits are self-referential and all that entails). And any non-equal sharing means the consensus becomes centralized over time according to the rule that the one with the most resources has the most economies-of-scale.

Than the circle is closed since they initially said it's not a ccy, but fuel.

The only way around this that I can see is to make Ether non fungible, which is a bit of a problem for a currency.

You got it with your edit.

Of course the virtual machine assumes they are atomic as you first stated, but the reality of the sharding is they can't be.

Building the virtual machine was the fun, easy part. The consensus scaling was the hard part which they should have done first even before raising ICO money. That is disgraceful.

They will be - the gas is paid inside the same transaction which invokes the script?

edit: but, of course that does invoke the horror of double spends - since partitions are necessarily separate and unaware of each other, I can spend the same gas in multiple partitions at the same time.

That doesn't matter (or actually it makes the problem I explained more obvious). The script validation and the gas spend confirmation must be atomic, which means they must be confirmed in the same partition, else validation can not be partitioned per the logic I already explained.

Gas is paid by the invoker of the script, you are aware of that?

In a strict sense you (and monsterer who mentioned that first) are correct, which is why I wrote the following:


Which I subsequently elaborated on today. The point about gas is why it is impossible to have sharded validation even if the scripts can otherwise be guaranteed to respect their partitioning.

But in a real world scenario where scripting is not controlled but is open for anyone to code, then when these scripts fail, they may not only fail themselves, but they may also do other things which cause other scripts to fail and even cause the economics of 51% attacks to be much easier and more plausible (make sure you click that link, as well read how I explained it to monsterer upthread). In short, the Nash equilibrium can collapse due to unpredictable externalities. But even if you argue against this, the sharding (paritioning) of gas is implausible with sharded validation.


Yeah that is my point that even though the virtual machine works, it doesn't scale decentralized (yeah sure we can scale anything centralized). And it won't even scale to where Bitcoin is now decentralized because scripts consume more validation resources than validating ECDSA signatures.

Here are some ideas for DApps to think about when designing a smart contract block chain:

www.coindesk.com/7-cool-decentralized-apps-built-ethereum/

I am of course thinking about this and how to do it correctly.

If I get healthy, I will try to teach these young guys a lesson about respecting those who have experience. I don't think Vitalik and Vlad are entirely clueless (and certainly Greg Meredith is a true academic and had worked at Microsoft Research), but I do believe they lack the experience in creating software and shipping it to millions of users. They lack some of the things I had to learn the hard way in those many years of shipping software that worked.

For example, I long ago shortcut directly to the generative essence of the problem space and thus am not wasting everyone's time and $millions pursuing braindead designs that can't possibly work. Instead I understand that validation must be centralized in any possible consensus system and thus I turned my attention to decentralized control over centralized validation. Those guys are instead still stuck back at figuring out what can't work.