Bitcoin Forum

Yesterday Yifu Guo (bitsyn.com (http://bitsyn.com)) and I visited a company that produces Android tablets.
We went with the intention of buying tablets, but when they asked how we wanted to use them,  it all started.

It became an intense meeting of nearly 5 hours. 
Nearly 75% of that time was spent discussing Bitcoins.
The managing director and other members of the meeting were very enthusiastic and seemed to get the big picture.
At one point,  he commented that "Bitcoin is a pandora's box."

The company is very interested in preloading ALL OF THEIR TABLETS with Bitcoin relay node software,  AND a Bitcoin wallet software.
They will then begin marketing their tablets as having this additional "Bitcoin feature" that their competitors don't.
They will market it along the lines that you can buy a tablet and become your own bank,  etc.

This company ships about 150,000 Android devices per month.
That is close to TWO MILLION tablets per year.

I think this is another huge opportunity, and giant step forward for Bitcoin.
I would suspect that with an extra 2M devices that have the Bitcoin relay node software always running in the background,
and that randomly connect to the internet from various IPs as their users use them, the Bitcoin network would become impossible to block.

Can someone with more technical knowledge chime in on the benefits of this many extra nodes?

In order for this to happen,  the Bitcoin community will need to be ready to help with the development of all the required software to do this.

Items required from us:
1) Super reliable Android Bitcion node relay software.
2) Super reliable Android Chinese Bitcoin wallet software.

They should consider making tablets that work as dedicated Bitcoin POS terminals for retail use.

This is epic on a scale that I have never imagined. Best of luck.

well that sounds pretty cool.

but i was never aware that there is a separate "relay node software"?  isn't that just built in to the current client?

I think preload is a good idea.

But preloading a full bitcoin node ?  I think diskspace & power-consuming will be problems.

I don't own a tablet, smartphone, or even a portable DVD player. I don't even like wall-phones, and I hate the idea of paying ~$80 for a phone contract when all I'd do is use it as a super-shitty netbook.

But... I'd buy this, if only to show support.  :o Keep us updated!

Right now they are one and the same, but there isn't any reason for them not to be separate. It just hasn't been coded. Might as well have full functionality though, IMHO.

what did the owner mean by the "pandora's box" comment?  sounds like he got over that pretty quick.

Bitcoin client on Android by Andreas Schildbach is a full client implementation, it should also include relay node functionality.

it "should" because it doesn't or "should" b/c you think it does?

Amazing news.

This has real potential.
 :o :o :o


I understood the Pandora's Box comment to mean something along the lines of "you can't put the toothpaste back into the tube" or "the cat's out of the bag" or "what has been seen can't be unseen," implying that once bitcoin reaches a certain threshold, there's no denying it's impact. (Perhaps we've already reached that point.) Is this what he meant?

Pretty awesome :) :)

Perhaps Piuk of Blockchain.info would build his android app with a special branded skin/UI for that company...

I also think it's a given in this discussion that the app would not be a full blockchain downloading app, but rather a thin client.

Also - perhaps each tablet should come with maybe 1btc or .1btc loaded in its balance already. The company would need to hedge prices by buying the bitcoins upfront.

It downloaded all block chain data into phone memory, I just not sure whether it also listen and relay unconfirmed transactions(may be not due to data and power usage consideration on mobile device). Need author clarification.

Yeah "pandora's box" tends to mean some combination of these three things:

1) Something with huge ramifications (good or bad)
2) Something dangerous
3) Something that can't be undone

I think Bitcoin fits all three definitions :)

This is exactly the reason Yifu and I went to visit them.
Yifu is working on this,  and is pretty far along from what I understand.
We were trying to source the final hardware supplier.


I think he was referring to #1 and #3 mostly,  and perhaps #2 in reference to the Chinese government.

By the way, it is open source. We can easily turn it into a full replay node if we don't need to consider data and power usage.

btw, i do run Schildbach on my Android but assumed it was a relay node b/c it holds the whole blockchain but i truly don't know...

The idea of having a tablet host a whole blockchain is sort of scary in and of itself.  Bitcoin explodes in popularity, and suddenly a tablet app will be non-functional without a hard drive to hold the block chain... and the sort of low-power CPU's in tablets are likely to have a difficult time keeping up with verifying a swarm of signatures, let alone bandwidth considerations.

That said, a tablet would make a fantastic dedicated device for accepting Bitcoins at retail, perhaps using a service that hosts the blockchain.  A built-in camera could scan the QR codes off of people's self-printed Bitcoin cash and immediately request a sweep of the funds.

I think back to my visit to Meze Grill... I got bitcoins, but the dude at the counter (not the owner) has no clue how to accept them.  He knows there's a QR code he can hold up and that I can pay it, but he has no independent way to confidently know that I paid with what he has there on the countertop, and asks me to pay with Visa.  If he had some gadget on his desktop telling him "$37.51 successfully received", the same way his Visa machine does, then it wouldn't have been so difficult.  A dedicated tablet could be just the niche for that purpose.

Awesome news Roger!!!!

Full node on every mobile device is not necessary and can't tackle the problem of "Great Firewall" blocking Bitcoin transactions. Another idea is setup full Bitcoin nodes in every major city with satellite internet connectivity.

This is awesome! Which tablet company is it?  ;D

1) Something with huge ramifications (good or bad)
2) Something dangerous
3) Something that can't be undone

I don't want to step on any toes,  so I will let Yifu check with the company to see if they are ready to have their intentions be publicly known.


Also,  I believe the idea is to install software in each tablet that relays new Bitcoin transactions throughout the network.
It would not store the entire blockchain.  I will need advice from the core developers in regards to what will be the most useful and practical.
If the Bitcoin software degrades the end user experience,  I don't think they will be willing to install it.

I'm not a developer, but I think I have a pretty good understanding under the hood to confidently suggest the following: don't bother having tablets be relays.  These are far better suited to being a client of some other more powerful server.  If there were a common language/API where such client could get services from any one of several providers, that might be way more attractive than having an app that depends on a single service that could go rogue, down, or otherwise change plans anytime.

An app that simply did "relaying" would simply consume the user's resources - in painful showstopper quantities - without adding any useful benefit to the Bitcoin network.

Without the block chain, the app can't validate transactions (so it doesn't know what's OK to relay and what's not), and it can't satisfy requests for historical blocks from peers trying to get their block chain for the first time.  Setting aside the fact that at a minimum, the entire subset of the block chain representing unspent transactions is needed to validate transactions, validating ECDSA signatures (to know whether they're worth relaying) is CPU-intensive enough by itself to matter enormously for a low-power device.

Consider, in comparison, a self-proclaimed "supernode" like Blockchain.info, which boasts "3356 nodes connected" at the bottom of their page.  If this is really a single Bitcoin node that is very well connected, this is the sort of thing that boosts propagation: the moment any transaction hits Blockchain.info, it will be instantly propagated to 3355 other nodes in a single hop.  The more nodes like this we can encourage to exist... this will give the P2P network the boost I think you are hoping for.

As Bitcoin grows and the transaction volume increases, unfortunately the kind of P2P network needed to sustain it will consist of highly-connected highly-powered nodes, rather than lots of low-power low-bandwidth ones.  This is just a simple matter of relatively undisputed fact, stemming from the current design of "all nodes must process all transactions".

If the company itself were to host a so called "supernode" and the clients were to always connect to their supernodes and another client I think this would give some nice connectivity.

This is pretty awsome. Pandoras box indeed.

like

Bitcoin Wallet for Android (https://play.google.com/store/apps/details?id=de.schildbach.wallet) does keep all block headers on the device. It dumps all transactions except the ones relevant for the wallet because of space constraints.

This means it cannot relay blocks right now. On devices with several GB of memory I'd like to see that happening in future.

If anyone wants to help coding, BitCoinJ (http://code.google.com/p/bitcoinj/) is the framework Bitcoin Wallet builds on and it would accept patches.

hehe... satalite...

We must get CyanogenMod and miui to include Bitcoin app by default

Yeah, this is a good idea :)

This is an interesting development!

For tablets it's clear that some sort of thin client will be best but so far the thin clients have been tied to individual servers or services. I think a network of utility servers and clients that could connect to them would improve upon this model. The tablet would hold(and backup) private keys and interact with one or more of these servers to get balance updates and relay transactions.

Has anyone started to develop such a server?

Before developing a server, a protocol standard needs to be agreed on. Otherwise, everyone will be doing its own protocol.

I'd like to remind everyone that it's pretty common for Chinese businessmen to make lofty promises like this. It doesn't actually mean anything in 99% of the cases.

Also, it would make sense to preload a tablet with a bitcoin client connected to a node that the company centrally maintains. But it's a terrible idea to have every tablet be relay nodes, not to mention impossible.

This is great!

For Android tablets, the BitcoinJ based application is probably the way to go. It doesn't store the entire blockchain, only what concerns its own wallet. Eventually the protocol could be evolved so that it doesn't even need to download entire blocks, only the chunks which contain the transactions it needs to know about.
Private keys are stored locally, no server needed.

But I guess some security features should be implemented before. Last time I've checked, I couldn't even access the wallet file in my android phone to back it up. Backups and encrypted wallets at least are a must. And even that is still vulnerable to malware, unfortunately. Maybe they should also consider making an agreement with some backup service and ship the table with their app included, like Wuala for example, which encrypts everything locally before sending the files to their servers.

We should probably organize bounties to the development of such needed features.

Why are you guys keep talking about BitcoinJ and etc... blockchain.info-like wallets are the future..

Did you ever see this thread:
https://bitcointalk.org/index.php?topic=7539.120

I know you programmed a POS terminal, but using QR codes.

What do you think of our/my design?
2$ a card.
Tops 100$ a terminal including your android phone in that price.

The card just signs transactions and gives them to the terminal which broadcasts them.
If they are invalid I assume they will just be ignored by the network.


If China starts using bitcoin it is game over. They have a lot of restrictions on moving money I believe so bitcoin really would be like pandoras box there.

Are you sure this is a good idea? I don't think a mobile device should run a node that will keep on consuming more ressources over time (cpu, battery, bandwith)

Encrypted wallets need to be implemented at BitCoinJ level. Note that you would not want to encrypt the whole wallet but just the private keys. Otherweise you'd need to enter your passphrase just to sync the blockchain and receive payments.

(Encrypted) backups I have implemented in a private branch of Bitcoin Wallet for some months now, but it is waiting for BitCoinJ to implement a reliable way to import keys/rewind blockchain, which is needed for restore.

So either way, patches for BitCoinJ (http://code.google.com/p/bitcoinj/) are badly needed.

BitcoinJ doesn't need a server. For Chinese people in particular, I guess not having all their transaction log stored in a particular place is a must.

+1

No way, I'd never give away my private keys to anyone.

By the way, the official blockchain.info app also uses BitCoinJ (-:

Bounty time?

I'm a Java developer, perhaps besides pledging I may also help with some little stuff that doesn't require deep knowledge of the code to be done quickly. I don't have much free time though, unfortunately.

Then use a client like BitcoinSpinner, which rely on a central server only for hanling the blockchain, but you and you alone keep the private key.

You don't need to give away your private keys. But the transaction history will be known by the server, unless you use their Alerts Disabled mode (https://blockchain.info/wallet/anonymity), but then I don't know how would your local software know how much money you've got.


Sure? I thought it was BitconJS, the javascript implementation.

First step would be to join the developer group (http://groups.google.com/group/bitcoinj). If you describe which feature you want to implement, I'm sure we'll help you with separating this into small steps.

imo, such a deployment should use some kind of intermediate bitcoin supernode. either something like the blockchain client or BCCAPI. for simple consumers of these devices it would be unacceptable to know anything about blockchains, loading times.

properly implemented this would have no real downsides. keys on device, passwords for encryption.

next step: ship the tablets with an app store that uses bitcoins for purchase!

Yes, because the blockchain.info app is a fork of an earlier version of Bitcoin Wallet.

https://github.com/zootreeves/My-Wallet-Android/blob/master/pom.xml
(search for the bitcoinj dependency)

Android uses Java, not JavaScript.

+1

BCCAPI allow lightweight client without having to trust a third party with his wallet keys.

Of course, what would also be awesome would be to be able to buy these tablets with bitcoins. And to have a market app accepting bitcoin :)

All implementations I've seen had a privacy downside. The server knows everything that goes in and out of your wallet.

I'm thinking, that could be avoided though.

Some steps that could be implemented so that the server cannot see the transactions of its clients:

• First of all, the server must not have access neither to the private, nor to the public keys of its clients.
• At each new block, the server sends it to the client, so that it can verify if it received anything.
• Alternatively, to decrease bandwidth usage, if the block has a large number of outputs (> 500?), the server can send only the output list of each block. If the client has an address on such list, it requests the entire block. The server will not be able to infer which addresses belong to the client if the list is large enough, and if people don't reuse their addresses much

It still possible to the server to log the outgoing transactions though. Maybe that could be avoided by requesting a list of bitcoin nodes to the server, and sending the transaction to these nodes directly instead of routing it through the server. But then these nodes could see the client's IP, if it doesn't use Tor or something else.

I wonder... is this the way blockchain.info Alerts Disabled mode works?

are you going to run full bitcoin nodes on those tablets?
are they for exportation or for the chinese domestic market?

Oh, OK, but I was thinking about the web app. That uses what?

It is necessary and can tackle the problem of the Great Firewall if these devices form an ad-hoc network (https://en.wikipedia.org/wiki/Wireless_ad-hoc_network) in addition to connecting to the Internet. Then as long as at least one device can connect to a free Internet, that node can relay everyone else's transactions. It's brilliant.

he, I guess I just described how a light weighted p2p client like BitcoinJ works. And plus, it doesn't rely on a single server that can be easily blocked by the Great Firewall.

Yeah, the more I think about it, the more I conclude that going p2p instead of client-server is imperative here. It's perfectly possible to do it "lightweight" to the point that even a mobile phone can handle. A tablet should take it easily.

It is not necessary (to be full node while being p2p) and I definitely wouldn't count on forming an ad-hoc network.

+1

I think we should think about reasonable enhancements to the existing Bitcoin P2P protocol and discuss that with the whole Bitcoin developer community (BIP process).

Goals to tackle:

- better support for lightweight clients
- being more firewall agnostic (perhaps by defining a HTTP(S)/SPDY variant of the protocol?)

Also, the peer discovery needs to be enhanced. Currently it uses DNS and IRC, and for both I think its a fixed server (that could be blocked easily).

This is truly awesome news, roger. The chinese are so
entrepreneureal once they "get" bitcoin the drive to leverage it for all sorts of things will be awesome - and lead to awesome things.

Not really.  An ad-hoc Bitcoin network would be a sitting duck for authorities to shut down at a firewall level if they didn't like it: Bitcoin network nodes promiscuously track and advertise their own IP's and the IP's of other nodes, giving the authorities a reliable targeted list of exactly where to find Bitcoin users and who to block to cut off the network, exactly the same way they can go find out the list of public Tor nodes from public Tor servers and block them in bulk.  Bitcoin traffic is also fairly easy to identify at a firewall level.

On the other hand, if a Bitcoin-behind-firewall setup is to be resilient, then the traffic needs to blend in and look like typical traffic for other non-banned applications.  It needs to look like SSL web browsing, or like instant messaging or Skype or BitTorrent at the network level, and it needs to go to a diverse set of IP's all around the world.

Given a company able to produce appliances, one nifty Bitcoin appliance they could produce is a router for non-Chinese customers that could earn Bitcoins for its owner.  That router would allow the non-Chinese user's IP to be borrowed for relaying traffic from China, to be administered by a centralized VPN company that leased the use of that router to a very small number of customers, who of course would pay in Bitcoin and the revenues would be shared with the router owners, much like a "Fonero".  With routers spread across a highly diverse set of IP's and no list of them published where the Great Firewall admins can block them all in a single brush stroke, THIS would bring resiliency to the ability to move Bitcoins through firewalls.

Why isn't piuk all over this already? Earth calling piuk... Earth calling piuk... This could be huge.

Ad-hoc networks cannot be controlled by authorities. The nodes themselves are the routers. The only way they could control such a thing would be to physically search each node, seize them and punish their operators somehow. That would be much more expensive than keeping a firewall, and would definitely increase the level of "tension" between the Chinese government and its subjects.
But the thing is, current technology and its dissemination is many years away from being able of support such a thing. We cannot rely on ad-hoc networks.

To give you some kind of info about what kind of problem this is: http://www.youtube.com/watch?v=DX46Qv_b7F4

But for starters, the developers need not to worry that much - just make simple lightweight clients or something, and if they really bundle their devices with those, awesome! It hink the chinese goverment won't even think about bitcoin for a long time...

Yes, exactly this!  Get it going ASAP, and if/when the Chinese government tries to block it, THEN work on solutions for getting around said block.

Ohhhh... I get it, he means "ad-hoc" in the Wi-Fi sense.

So all of the Bitcoin nodes in the whole country have to be within a couple hundred feet of one another.

I know China is population dense and all, but I am not convinced that would work... if it was that magic, all of Africa would be online with it by now. China seemingly has no qualms about punishing operators of things they don't like... they just... disappear and become organ donors, or human body exhibits at the Luxor in Las Vegas.

I don't think anyone here is giving the Chinese government much credit. There are alternatives other than the PRC clamping down on Bitcoin. If there are enough bright men and women in the lower echelons of the government, who are capable of understanding Bitcoin and its implications, it could quite easily turn into China either accepting Bitcoin or co-opting it.

If you think about it, a country that tends to keep its own currency artificially deflated could very well see Bitcoin as a way to offset the pitfalls of an artificially deflated currency. Keep the Yuan for intra-Chinese commerce, adopt Bitcoin for international commerce and profit from the spread between them.

You have to figure the entire country cannot possibly be so stupid as to have knee-jerk reactions. Someone, somewhere, is looking long and hard at this and trying to figure out what the best course of action would be.

I'm happy to discuss the technicalities of this with you guys. You can join the bitcoinj list mailing list here:

  https://groups.google.com/forum/?fromgroups#!forum/bitcoinj

or we can discuss here too, though it's a bit less convenient.

Here's a quick summary of the possible approaches:

• Satoshis code relaying transactions. Not going to happen on a power / memory / storage constrained device like a tablet. We can safely put this to one side.
• A "simplified payment verification" (SPV) client, like anything bitcoinj based such as Android Wallet. This doesn't do a whole lot for the network but it allows people to make/receive payments with their tablet, which could make a lot of sense for point-of-sale applications. It has the advantage of not having any central servers that can be blocked with a simple firewall rule, and the privacy is good because you just use the regular P2P network. However it does still have some delays and is more resource intensive than the third option. And some modern features people want like deterministic encrypted wallets aren't available. Over time these problems can be fixed but with todays code, that isn't the case.
• A superlightweight client like BitcoinSpinner. This has the advantage of being instant-on and quite robust on the client side. The problem from the China perspective is that the server (and thus the governments) knows all your keys, can therefore connect all your transactions, and can also block connections to the remote server breaking the app. Whilst these clients can in theory be pointed at different servers, you have to find such a server, decide you trust the operator, and reconfigure your client. In practice most users wouldn't do that. They'd just assume the app was broken and uninstall it.

My preference is obviously for the bitcoinj SPV approach design-wise, but it does have risks. The code develops slowly as I don't have much time to work on it, and in general fitting the pure P2P approach onto a phone or tablet involves some solving some tricky data structure and optimization problems. Long term it'll end up the best approach IMHO, but in the shorter term something like piuks wallet may work better. It boils down to timelines.

Bear in mind just being preloaded onto tablets, does not mean anyone will use the apps. There are still many problems none of these apps solve:   how do you educate users about Bitcoin? How do you make it easy to buy small amounts without setting up an account at an exchange? How do you help users find each other so they can trade?

The amount of work here is just enormous, and I'd like to see as much of it as possible be shared between client developers regardless of which protocol design they choose.

I will defer to people with more knowledge in regards to what kind of software should be installed.

The majority of their tablets are sold in the domestic Chinese market,  but they ship a substantial amount world wide.

From current solutions, I would go with BitcoinSpinner. While other clients are OK, waiting for anything like block chain updates are jsut pain in the ass for average Joe, and will amke them hate bitcoin.

And I would strongly recommend something more like Bitcoin Wallet for Android since the blockchain doesn't get stored and it can't be so simply blocked/shut down. The next step should be building more servers like the one powering BitcoinSpinner(at least for markets where the government is assumed to get involved as soon as they hear about it).

I see. Note that Electrum could be used.
It is now available for Android, although the Android gui is not as advanced as the Qt version.

It's not magic. The few nodes that are connected to the Internet have to pay the bandwidth costs of the entire network, which for most Internet services increases in proportion to the size of the ad-hoc network. But, and this is the cool part, the Internet bandwidth used by a (full) Bitcoin node does not depend on the size of the ad-hoc network - you still download the same number of blocks and relay the same number of transactions regardless of how many ad-hoc peers you have. Each node could connect to hundreds or possibly even thousands of other nodes on the ad-hoc network (which doesn't cost anything and the only limit is how much bandwidth the hardware can physically handle) while simultaneously attempting to connect to a handful of nodes on the Internet (if and when a Bitcoin-friendly Internet connection becomes available).

I'm not 100% certain it would work either (as you say, each node needs to be in close proximity to another node), but this is exactly the sort of application that ad-hoc networks are good at, so I think it has a real chance of success. Remember, the ad-hoc networking is in addition to the Internet connection, not instead of it (and in fact it's completely useless without it), so even if the ad-hoc networking is a complete failure, it still functions as a regular relay node.

afaik, large-scale mesh ad-hoc networking on mobile devices is still an unsolved problem.

Nice, but there is a small problem. Everyday a new flaw MAY be discovered within bitcoin software, I can't imagine 150k people updating their tablets just to get the new client update(assuming they ALL use Bitcoin).

Why not?  Updates to software happen all the time.

Work in progress... the Serval Project (http://www.servalproject.org/) is a project with the aim of implementing a mesh network targeting Africa.

This was a cool thread. Is any of this beyond the vaporware stage?

Definitely a game changer imho

IMHO, i think that btc clients should be able to operate behind tor.

this would remove many of the difficulties related to the Chinese communicating that have been talked about in this thread.

Agreed

I think Rogers had 5 hrs of beers and drugs.... so they can all talk vaporware.

Any chance this company is the shitty  "COBY" brand?

I sure hope Rogers didnt pay for beers if it is.

having a full node running in the background for users that don't even know about bitcoin is not a good thing.
imagine microsoft re-develop their onlive service so that each user of windows is part of their network much like a torrent system so that everyone was whether they want it or not was getting things downloaded and uploaded from their PC.

BLOATWARE / Malware / ADWARE comes to mind.

i would simply have a light app that allowed people when clicking on it to act like an android app that talks to a blockchain.info wallet or something similar.



no secret/always running in background, no bloated taking up gigabytes of hard drive, no hours of slow internet and computer sluggishness while it downloads the blockchain.

something that PiUK should definitely get involved in

a simple and light wallet with a nice small presentation/tutorial/ introduction to bitcoin.

This is a great idea.  Blockchain.info already has Chinese as a language option.