What is the ideal hardware configuration to become a B&C reputed signer?

[Sabreiib]

It is said that Tor has less connection width, and I believe we should distribute risk by thousands random signers rather than a dozen of famous singers.

[Nagalim]

Ideally customers will have a choice of really large multisig with many lower trust or small with high trust. Signers can go from low trust to high trust by first participating reliably in super large multisigs. The limit will be the blockchain tech with higher fess and limiting n-of-m configurations.

[JordanLee]

@JordanLee Do you have an opinion on the safety of operating Bitcoin wallets over Tor? At least one researcher indicates that it makes the transactions less safe and more prone to theft: https://news.ycombinator.com/item?id=8497303

It is true that man in the middle attacks can be performed by a Tor exit node. That can be a problem when sending unencrypted data that can be tampered with. Messages/transactions sent over the B&C Exchange and Bitcoin networks are indeed unencrypted, but the signature makes them tamper proof. A Tor exit node has the same options for dealing with a Bitcoin transaction as any Bitcoin client. Theoretically they could decide not to transmit a transaction just as a Bitcoin client could choose to censor transactions, but they don’t have any motive to censor Bitcoin transactions. Therefore, I don’t think it will happen.

I don’t see how running a Bitcoin or B&C Exchange client over Tor could place any funds at risk.

[JordanLee]

Is it an option to limit the amounts you sign for? In that case a signer only signing e.g. under $1,000 may still opt to use a VPN as the risk is relatively limited. Still trying to understand how the selection of signers is made by the traders and whether signers have an option to sign for limited amounts as advertised.

Sorry, signers cannot limit the quantity of money they sign. The system has a max trade size for each asset, but this is a network wide variable set by shareholders.

From the design paper about how signers are chosen:

To construct this [a deposit address request], a client must first determine which reputed signers it will use, which will be the ones with the highest reputation score. Once these signing candidates are determined, the blockchain must be searched for a deposit address list published by a particular signer which matches the asset ID of the deposit address request. An address must be confirmed to have not been used before, which means it does not appear in any other deposit address request on the blockchain. If a reputed signing candidate does not have an eligible and valid address the requester can use, another signer should be chosen (the one with the next best reputation).

[JordanLee]

I believe we should distribute risk by thousands random signers rather than a dozen of famous singer

The Bitcoin protocol and most other protocols support a maximum of 15 signers. We can change what is possible in the B&C and Nu protocols, but each signer adds to the size of the message.

[tomjoad]

It is said that Tor has less connection width, and I believe we should distribute risk by thousands random signers rather than a dozen of famous singers.

More signers arguably increases the risk, if there is no burden or cost to becoming one. For example, a botnet operator could “create” thousands of signing identities and control the signing power of the network. How would the network differentiate between honest and dishonest signers at that scale?

There must be a cost associated with becoming a signer to maintain quality. In our current setup, that cost is reputational, which presumably must be curated over time by a signer through forum postings and community interaction. A signer with no reputation will not be very valuable to the B&C Exchange network.

[Sabreiib]

The maximum of 15 multisig signers is OK, I suggest we set up a pool, perhaps there are 1000 signers in this pool. And customers have the options to pick 2-15 signers from 1000 signers. The software’s default selection is the top signers with most reputation vote, but customers can change it by manually pick or random pick automatically.

E.g. We may set up 3 class pools, the 1st pool is only 15 top signers and can handle any amount of transactions.The 2nd class pool with 150 signers handling 1-1000 NBT value transaction, the 3rd pool with 1500 signers can only perform micropayment(below 1NBT).

I’m afraid if we have only a dozen of signers, the criminal or government may locate them one by one, and destroy the whole system. Tor is not 100% safe, plz refer to silk road.

[Sabreiib]

The cost is BKS itself, the most well known PoS anti-attack machanism.

I won’t vote BKS addresses with insufficient balance, if I find these address get voted, I will downward vote them! In this stage, I will up vote you @tomjoad, because you are like satoshi in BTC network in 2009. But in the long run, I only vote big-medium BKS holders.

And for myself, DON’T TRUST ME, I am not sure I won’t become corrupted in future or threatened by someone when typing words on forum, of course you can agree with the logic I said and mathematics. If you really wanna trust something in human nature, then believe in the “selfishness”. Because selfishness is one of the most stable featrure of human being and has been proved for thousands years. You can belive BKS holders will do their best to maximize their interest by promote B&C DAO.

[Nagalim]

To put some math to it, assume each signer has a reputation x. We can rank a multisig by the total reputation required to activate its funds. So for a y-of-z address there is a maximum possible reputation for a transaction equal to the sum[x_i for i from 1 to y] for the most reputable signers. With a given reputation bar, the multisig could be 15-of-15 or 1-of-1 and be of similar strength. From there we can introduce the concept of the lowest reputable txn possible and minimize that value while attempting to reduce response time by lowering the bar for consensus (from 100%, like 15-of-15).

[zoro]

selfishness

you mean profit? businesses? entrepreneurship?
If yes, i agree. Nu,B&C and the like cannot rely only οn volunteers!

[Sabreiib]

[Sabreiib]

Back to the topic. Intel NUC should be the ideal hardware for signers.

http://www.anandtech.com/show/10121/intel-nuc6i5syk-skylake-ucff-pc-review

[Cybnate]

Intel NUC should be the ideal hardware for signers.

Depends on your budget, existing hardware and availability of reliable internet I suppose.

[masterOfDisaster]

To host the blockchains this might be sufficient:

ZOTAC releases ZBOX BI323 and CI323 nano Mini PCs

(speaking of the CI323 in particular).

It’s quite cheap, silent, energy efficient and the quad core Atom processor should be fast enough to keep some blockchains, for which the signer offers service, synced.

The BCE blockchain will be hosted on a RaspberryPi3 that also runs the BCE application and will be used as “signing environment”.

At least that’s the setup I have in mind.

Backing up each of the devices should be fairly simple with this setup as well as having standby devices ready in case the “blockchain machine” or the “signing machine” fails.

I’m still not 100% sure whether it’s clever or not to split the signing environment from the rest, but I have in mind that it makes it easier to apply hardening to the signing machine.

This setup could use “blockchain machines” on VPS or other devices.
You can easily create dedicated VPS per blockchain (one for BTC, one for LTC, etc.).
There’s so many options, if you split the BCE stuff from the rest…

[willy]

I’m eager to hear about your experiences.

I ordered three of those yesterday:

http://www.banana-pi.org/m3.html

Should be enough for seven daemons (all supported coins except NXT + BTS) so far.
Let’s see what the future brings and how many coins B&C will support.

[zoro]

A single laptop with a lot of ram and SSD will be much efficient holding a lot of coins

[masterOfDisaster]

The specs of the BananaPi sound impressive!
I suppose you intend to run the blockchains of BTC etc. as well as BCE on such a device?
2 GB RAM should be a little bit too low to run more than, say 2 or 3 daemons at one such device.

That’s why I was looking for a quite cheap solution with a lot of RAM (and hopefully sufficient CPU power) and ended up considering 3150 Atom boards.
8 GB RAM should be plenty.

ASRock Beebox even supports 16 GB RAM:
http://www.anandtech.com/show/9434/asrock-beebox-review-a-fanless-braswell-ucff-pc

I’m still on the fence what to try…

A single laptop with a lot of ram and SSD will be much efficient holding a lot of coins

If you can save the display etc. and get a device for a little over $200 (including 8 GB, 500 GB HDD), this will beat the laptop approach (you don’t need SSD speed for that blockchain stuff…).

[willy]

2 GB RAM should be a little bit too low to run more than, say 2 or 3 daemons at one such device.

Yes, that’s my main concern.

I have a test setup with all the 7 daemons running… the hungriest is nud. All daemons together use around 4 GB.
BTC, LTC, PPC and DASH are pretty lightweight.
DOGE and B&C are okay for now, Nu 2.0.3 … well you all know.

As for the barebone solutions, I’m not happy with the number of cores / threads of the CPUs. Those are build for different purposes.

[masterOfDisaster]

hungriest is nud

nud 2.1.1 is way more efficient!
I suppose you speak of 2.0.3, right?

[willy]

2.0.3, right?

Nu 2.0.3 … well you all know.

yes