Vitalik Buterin compares bitshares vs ethereum like ASIC vs CPU. Thoughts?

[JoeyD]

There does seem to be a market for the term "Turing-complete" in _the crypto market_. A large part of the market seems to think that all other cryptos, must be "incomplete" and rather have them some of that "complete"-stuff, somewhat similar to the bitcoin-2.0 thing. I wonder what version-number the internet is supposed to be now, is it time to upgrade my e-mail?

Nevertheless it does seem to be a rather clever marketing move to use the of lack of understanding of the crypto-masses. I doubt that calling themselves the java-vm of the crypto-world, would have garnered them the same attention. However I personally don't feel very comfortable when such a method of marketing is being used and I hope bitshares and I3 will use other methods.

[xeroc]

Let us not forget that the ASICs are dominating the market 

... of bitcoin MINING ..

take a look at other PoW schemes (ie, primecoin, gridcoin, quarkcoin, ..... lots more) or even PoS coins (ie, nxt, peercoin, mintcoin, ... lots more)

there is no _the market_ in crypto

[sschechter]

Let us not forget that the ASICs are dominating the market 

[JoeyD]

I had trouble making out all that was said in that video, the audio was terrible on my laptop.
I also tried listening to it without any prejudice and there were a couple of things I noticed with regards to comparisons of Ethereum to Bitshares.

If I heard correctly the first part seems to be an attempt to describe ways on how to recreate something like bitsharesX on top of Ethereum. However the way Vitalik described it and the words I could catch, I was unable to figure out how his version was easier or more efficient in anyway (the requirement of external feeds also did not sound very safe). To be perfectly honest his proposed solution sounded like the exact same thing as Bytemaster is doing now, but with an extra layer of complexity, uncertainty and inefficiency, which leaves me questioning if I missed something.

Correct me if I'm wrong, but did they explicitly say that not all essential parts for Ethereum to work were solved or even figured out and that they hoped for solutions via bounties or workgroups? Did they also say that for those solutions to be implemented they absolutely need a complete replacement and migration in the form of Ethereum2.0 (or more) by design? Should those things not be solved before launch and presale? Wouldn't such a design be disastrous for all the separate DACs on top of Ethereum? To me that made the whole proposition sound dangerously interdependent and fragile and I don't understand how their solution would be preferable to having independent separate DACs. At the very least it did not sound like a very strong argument to build any serious DAC on top of their system.

Talking about computer analogies that made their proposition sound an awful lot like having to periodically reinstall your entire OS plus all applications and then hope everything still works as expected.

It's probably because I'm to stupid, but after watching that video I'm having more trouble than ever seeing how their solution is any more elegant or efficient or "general purpose" than separate alt-coins, merged mining or bitshares. Calling Bitshares Asics made their solution sound like simulating asics on their cpu. I'm starting to suspect that their ambitions are too high/broad (feature-creep?) and that their focus should be more on creating a more elaborate scripting engine than is currently available in Bitcoin.

EDIT
Missed a couple of post above this one, that seem to answer some of my questions. Stans description was especially to the point and well worded.

[toast]

I think Vitalik's original point here just shows the weakness of trying to explain things too much in the form of analogies. While it can be helpful, such analogies often lead people to false conclusions when people try to attribute all kinds of characteristics of the analogous system to the actual system. In practice, there's very little involved in either design or execution of a BitShares DAC that is similar to the design or operation of an ASIC (I can say this with confidence, having been involved in both software and FPGA/ASIC design for most of my career).

Also, I suppose one of the ideas being promoted by the original analogy is that the logic of a bitshares DAC isn't changeable once a blockchain is started (ASICs are, to a large extent, fixed function devices), but there's nothing in the Bitshares technology that prevents changes in DAC operation based either upon agreement by individuals or by groups. We're just not explicitly building such out-of-the-box capability into the initial "DAC toolkit". Individual DACs could certainly build in "runtime-changeable" parameters.

Stan's analogy is actually by far the closest comparison, in my opinion, based on what I've heard about Ethereum so far. As I understand it (and I confess I haven't read about it depth, because the initial concept just didn't seem efficient to me), Ethereum is basically going to be running a scripting language interpreter on each peer that processes script commands in a blockchain block.

 

[dannotestein]

I think Vitalik's original point here just shows the weakness of trying to explain things too much in the form of analogies. While it can be helpful, such analogies often lead people to false conclusions when people try to attribute all kinds of characteristics of the analogous system to the actual system. In practice, there's very little involved in either design or execution of a BitShares DAC that is similar to the design or operation of an ASIC (I can say this with confidence, having been involved in both software and FPGA/ASIC design for most of my career).

Also, I suppose one of the ideas being promoted by the original analogy is that the logic of a bitshares DAC isn't changeable once a blockchain is started (ASICs are, to a large extent, fixed function devices), but there's nothing in the Bitshares technology that prevents changes in DAC operation based either upon agreement by individuals or by groups. We're just not explicitly building such out-of-the-box capability into the initial "DAC toolkit". Individual DACs could certainly build in "runtime-changeable" parameters.

Stan's analogy is actually by far the closest comparison, in my opinion, based on what I've heard about Ethereum so far. As I understand it (and I confess I haven't read about it depth, because the initial concept just didn't seem efficient to me), Ethereum is basically going to be running a scripting language interpreter on each peer that processes script commands in a blockchain block.

[clout]

Bitshares is more like a RISC processor while Etherium is more like CISC.
RISC is more efficient. ARM is RISC for example. Power efficient, low cost, highly specialized.

i think that the risc vs cisc analogy is most applicable.

I think it is more CENTRAL Processing Unit  vs   GPU... think about it like this:

1) CPU can do anything, but it is centralized and doesn't do graphics well
2) GPU is DATA parallel (multiple chains) and application specific (optimized for graphics)

When viewed from the perspective of a DAC... efficiency effects profitability because bandwidth, storage, and CPU costs are paid by the network.

hmmm.... i obviously think that this is a good analogy as well but i think the same points can be made when referring to risc and cisc. not to mention the cost efficiency is more apparent with the risc architecture and is why it used predominantly in smartphones and wearable technologies. in the same way that I look at risc dominating the market in the future because much of the third worlds introduction to computers and the internet will be through smartphones and tablets, I see people in the third world using bitshares dacs before using ethereum because their computational resources will only allow them to use the most light weight and efficient software.

[bytemaster]

Bitshares is more like a RISC processor while Etherium is more like CISC.
RISC is more efficient. ARM is RISC for example. Power efficient, low cost, highly specialized.

i think that the risc vs cisc analogy is most applicable.

I think it is more CENTRAL Processing Unit  vs   GPU... think about it like this:

1) CPU can do anything, but it is centralized and doesn't do graphics well
2) GPU is DATA parallel (multiple chains) and application specific (optimized for graphics)

When viewed from the perspective of a DAC... efficiency effects profitability because bandwidth, storage, and CPU costs are paid by the network.   

[clout]

Bitshares is more like a RISC processor while Etherium is more like CISC.
RISC is more efficient. ARM is RISC for example. Power efficient, low cost, highly specialized.

i think that the risc vs cisc analogy is most applicable.

[Stan]

The problem with this analogy is that BitShares uses C++ which is Turing Complete and in no way a "reduced instruction set".

Perhaps a better example is comparing interpreted languages (BASIC, JAVA) to compiled languages (JAVA, C++).

Here the advantage is C++ being able to compile once and execute efficiently forever vs. interpreters doing the compiler's job at run time over and over again, forever.

Interpreters are more flexible for making up your mind what to do at run time, but if you're a BitShares Sovereign DAC, you already know exactly what you're going to do at run time.

 

[xeroc]

Isn't it kind of _nice_ to compare blockchain protocols with CPUs ander hardware :-) I like it :-)

[luckybit]

In some sense he is right .. ethereum is more like an Operating system on which you can run applications ... much like TCP to HTML/SMTP/...
Or like the BIOS is compared to Linux ..

Anyway .. as financial instrument bitshares are close to beeing developed .. whereas ethereum first needs to establish the basis operating system before others can create financal instruments (ie trading platforms)

Bitshares is already there NOT depending on a basic OS ... that's actually an advantage ..

Bitshares is more like a RISC processor while Etherium is more like CISC.
RISC is more efficient. ARM is RISC for example. Power efficient, low cost, highly specialized.

He is right that Bitshares chains are highly specialized but the benefits is that it's also more decentralized and more likely to be used and last for decades.

One chain to rule them all isn't a good idea and won't last even if it is turing complete. The reason is Ethereum is going to be forked almost immediately and then the what? Do they plan on it being forked?

So if it is forked all the investors, the miners? There is a good chance Ethereum will be forked and a better version of it will be released without some of the unpopular decisions built into it. This is going to happen just as Mastercoin has to deal with Counterparty.

For that reason, Bitshares has an advantage because Bitshares gets better if people fork it.  The one thing Ethereum team is better at is marketing. They have the miners on their side and they are winning politically. This is also happening because Bitshares team isn't doing enough to reach out to groups not associated with or aligned with the political views of the inner circle. This is necessary though if you want mainstream adoption because most people just want to know if it makes money or what it can do for them, the politics can be tailored and calibrated to the different communities on a chain by chain basis.

Ethereum team has one major weakness though, they appear as the insiders. They appear as the elite, even if this isn't actually the case. This is a failure of marketing on their side because they want to look as if they are with the little guy but giving conferences at Silicon Valley is not where you would find any little guys.

[davidpbrown]

Well that is the multi-million dollar question..

To continue the analogue above, why have the CPUs doing work that speicialist GPUs can do? Tying it all together is no mean feat though, so if they pull it off, fair reward due. ATM we don't know that it's not possible and in all likelihood it is practical in a useful way.

[CLains]

If Bitcoin core devs are arguing with Counterparty/Mastercoin, then surely the various applications on top of Ethereum are going to be arguing in just the same way, unless they go with parallel chains. Eventually they will figure out how to do parallel chains in a neat way, but how long will this all take, and will their financial model support it?

Buterin mentioned off-chain trading and merkle trees (?), ... Can someone explain in simple terms what he's on about, and how this might make parallel chain trading possible?

[davidpbrown]

I like this analogue. Obviously, "Application-Specific" tools are going to be easier to implement and better in the short to medium term.

Ethereum is ambitious to say the least and setting that up is going to take a lot of work. It will be interesting to see Ethereum develop to a point that it can draw on 'ASIC' like tools, such as BitShares, and then do more with those. I can't see Ethereum every replacing BitShares or others - even it's pretentions of doing Namecoin in 9 lines, understates the benefits of having a tool for the job. Ethereum might add flexibility but when you don't need flexibility, that overhead perhaps not needed either.