I had organized the clients into a similar table a few months ago, with something simple designed for the first-time user.
http://lovebitcoins.org/getStarted.html
I wanted to put more emphasis on the mobile clients, because this is much easier and faster for a first-time user to get up and running. This is especially true whenever I meet someone in person, since they always have their phone, but they never have a computer.
If the newbie likes what they see in the mobile, they can always become more of a power user.
Topic: New bitcoin.org Clients page - page 2. (Read 6051 times)
I went with the objective comparison table intentionally to avoid all this subjective arguing 
I think that your table is very useful, but it serves a different purpose: it is an objective comparison, not an introduction text.
For the 'clients' page, Amir asked me to write a 300 words description of Electrum, which I did.
Amir's request made me believe that the 'clients' page was going to contain descriptions of their work made by developers, and not an objective comparison by independent reviewers.
Amir brought some corrections to my text, which was consistent with the idea of developers describing their work, since he contributes to Electrum too.
But finally my text went to the trash, and Electrum gets a completely different 'review' text written by someone who probably never used it.
You can understand that I'm pissed.
why was I asked to write a description in the first place?
For instance Electrum has an android client - would that get another entry for mobile clients?
Sorry again Thomas, but I'd suggest not including Electrum on a page for Android clients. Here is what you arrive at if you click the "New Android client!" link:
http://ecdsa.org/electrum/android.html
Well presumably that will change in the future. I think it's best to have 2 sections - 'normal' clients and 'web wallets'. Mobile clients can be bunched informally together and we can use the screenshots, platform icons and description to distinguish them. Really mobile clients are already a grey area what with touchpads running android and/or ubuntu. I kinda see them merging in the future.
The only useful main distinction is whether you keep your own wallet or trust someone else. External services are always vulnerable to shut down. The Electrum-model is not a worry as several servers can be combined together, and they never have access to your wallet. The only worry is a lack of privacy which we enumerated in the description.
Yeah, but that doesn't remove the complexity, it just pushes it onto our users, who are least qualified to make a decision. Easier for us, not so great for wider adoption.
Though it's sucky it's probably better for us to just thrash it out here. The descriptions aren't set in stone. We could put something about not supporting the network on other clients, or better, just drop that part and make it clear that by running Bitcoin-Qt you are helping support the network and project, then just not mention it for the others.
Though it's sucky it's probably better for us to just thrash it out here. The descriptions aren't set in stone. We could put something about not supporting the network on other clients, or better, just drop that part and make it clear that by running Bitcoin-Qt you are helping support the network and project, then just not mention it for the others.
I went with the objective comparison table intentionally to avoid all this subjective arguing
For instance Electrum has an android client - would that get another entry for mobile clients?
Sorry again Thomas, but I'd suggest not including Electrum on a page for Android clients. Here is what you arrive at if you click the "New Android client!" link:
http://ecdsa.org/electrum/android.html
You are told to follow a 4 step process that involves manually download and install various packages that are not in the Market, download and manually run scripts, etc. This is by no reasonable standard describable as "very simple". Very simple means you go to the market web page and click install:
https://play.google.com/store/apps/details?id=de.schildbach.wallet&feature=search_result#?t=W251bGwsMSwxLDEsImRlLnNjaGlsZGJhY2gud2FsbGV0Il0.
then you're done. Upgrades happen automatically.
The given rationale for this convoluted install process is "this gives users the possibility to see what the code is doing, and to check that it does not contain malware". The issue is, again, Electrums website conflates users and developers. They are not the same people. Users cannot read Python and even many developers won't read it well enough to be sure there are no security issues lurking there, assuming they even bother.
If we actually suggested users follow this procedure we'd simply lose a lot of credibility. Again, the software might work very well, but there would need to be improvements to the installation / upgrade process.
As the main developer of Electrum, I strongly disagree with this message. It is inaccurate, and it sounds like you are trying to scare users off.
Electrum simplifies the use of Bitcoin, because it removes the need to download the complete blockchain, and to do regular backups of your wallet.
These two aspects of the Satoshi client make it require a substantially higher "level of technical ability" than Electrum.
Electrum simplifies the use of Bitcoin, because it removes the need to download the complete blockchain, and to do regular backups of your wallet.
These two aspects of the Satoshi client make it require a substantially higher "level of technical ability" than Electrum.
One issue with trying to sum up the differences between clients like this is that we'll inevitably run into disputes. But we don't have much choice - there has to be some guidance to users on how to pick which clients to try.
The reason I put this (and genjix agreed with me) is that your clients website is clearly not intended for regular end users. It lists "easy to review the source" as a feature, for example. It's great that you've focused on making backups easy and this is recognized in the text. That is only one component of usability. If I pointed somebody who is not a programmer to your page, or even someone who is but who isn't very experienced, they'd probably run into problems at the first step of simply installing the app. When was the last time you went to the website of a typical software company and was told to download/install Python yourself?
If the install and website was more end user focused, eg, provided regular installers for the common platforms, dropped the technical stuff from the front page, then I'd probably not have said that. It can certainly be changed in future. The core software itself doesn't seem to have any particular usability issues.
Anyway, the description isn't meant to be an exhaustive list of all features provided. It tries to pick a few distinguishing features and mentions them. The main distinguishing feature of Electrum is probably speed and the deterministic wallets, and those are mentioned (well, easy backups are mentioned).
Yes, if I wrote a description for a web wallet, I would mention the lack of privacy and potential security issues too.
Quote
So, the main question is whether the description of clients should be left to developers or independent reviewers.
Like I said, I haven't developed any of these clients. I wrote a library that is used by one of them, but that's about it. Amir also hasn't developed any clients, just a (different) library.
You're welcome to suggest alternative reviewers, but you'll note that I dished out some criticism of the client based on my library as well.
Next up would be a page for mobile / web wallets, maybe?
+1I think also that there must be a page/section dedicated to web wallets and another one to mobile clients.
Yep. 'Web Wallets' sounds good.
I'm not sure about a mobile section since that can be represented using the platform icons; for instance Electrum has an android client - would that get another entry for mobile clients?
ThomasV, sorry, but I agree with Mike Hearn text, saying explicitly that a client-server solution provides less privacy.
I was not complaining about that; I was complaining about what he wrote on the need to be "tech savvy".
Now if there's going to be a similar page about "web wallets", I hope that mike will be fair enough to mention the same lack of privacy associated with them, and the fact that "you don't contribute your computer's resources to the network" when using them.
Oh, and did I mention that Electrum too has "the ability to construct transactions whils disconnected from the internet"? no, I did not, because I think that is the sort of technical detail that we do not want to mention on that kind of page. but apparently Mike has decided it's important to be mentioned for some clients.
Noob users want "the Bitcoin client", they don't know what is the Satoshi client 
These "noob users" will probably read some recommendation somewhere, or receive it from a friend, of what client to download. How many users of uTorrent have ever accessed bittorrent.org?
We are not trying to sell something on bitcoin.org. This is "the protocol page", and it would better have a good level of transparency. If that implies some people will have to do a minimal homework before starting to use the software, so be it.
Also, I'm not even sure we should recommend bitcoin to "total noobs". The chances they end up losing their money to malware are too high, IMHO. "Total noobs" will have to wait for a dedicated, air gapped bitcoin device, unfortunately.
I liked Luke-Jr layout. Maybe a 'comparison' page with that table would be a good idea.
ThomasV, sorry, but I agree with Mike Hearn text, saying explicitly that a client-server solution provides less privacy. This is a fact, and users should at least be aware of it. That said, not sure the "more suited to tech-saavy" part is necessary...
Finally, Bitcoin-QT should definitely not be labeled "official". Even the term "original code from Satoshi" is not really accurate. AFAIK, only the bitcoind part is "original", the GUI is totally new. And Bitcoin-QT is not the only GUI using bitcoind, is it? (If I remember well Armory uses bitcoind too, making it as "original" as Bitcoin-QT. Please correct me if I'm wrong).
I appreciate the listing of multiple alternatives in bitcoin.org. Good job!
ThomasV, sorry, but I agree with Mike Hearn text, saying explicitly that a client-server solution provides less privacy. This is a fact, and users should at least be aware of it. That said, not sure the "more suited to tech-saavy" part is necessary...
Finally, Bitcoin-QT should definitely not be labeled "official". Even the term "original code from Satoshi" is not really accurate. AFAIK, only the bitcoind part is "original", the GUI is totally new. And Bitcoin-QT is not the only GUI using bitcoind, is it? (If I remember well Armory uses bitcoind too, making it as "original" as Bitcoin-QT. Please correct me if I'm wrong).
I appreciate the listing of multiple alternatives in bitcoin.org. Good job!
Noob users want "the Bitcoin client", they don't know what is the Satoshi client
Anyway you can add it as surname, ex:
Bitcoin client
( Satoshi client )
As I already said I also like to have a 'comparison' page
Anyway you can add it as surname, ex:
Bitcoin client
( Satoshi client )
As I already said I also like to have a 'comparison' page
I agree that for now, we should recommend Bitcoin-Qt as the default option until we have more experience with other clients and their general level of polish and robustness is better. Bitcoin-Qt has an unfortunate name. The UI toolkit used should really not be exposed to end users. I'd support renaming it to something like "Bitcoin Core" or "Bitcoin Classic".
+.1 What about "Satoshi Classic"?
Next up would be a page for mobile / web wallets, maybe?
+1I think also that there must be a page/section dedicated to web wallets and another one to mobile clients.
Looks good to me, thanks!
Next up would be a page for mobile / web wallets, maybe?
Next up would be a page for mobile / web wallets, maybe?
Thanks Mike, I updated the page with your descriptions. Although I do agree that the Electrum text is inaccurate - it's not user friendly *yet*, not because its core focus is technical people. I did move MultiBit first because your reasons were solid. I can agree that Electrum has a problem with distribution (no packages, difficult/obscure install instructions for most, random win32 builds from forum). Armory is not much better in that regards, so I used a coinflip to select their order
If anyone wants to test, then add this to your hosts file:
176.31.24.241 bitcoin.org
For all the complaining about centralisation on this forum, people sure seem to be fighting my efforts to promote alternative clients, to avoid promoting a strong exchange competitor by continuing to trade on MtGox and to avoid getting involved with development (bitcoin library with Python bindings is out there). We should not be pushing something because it is simply the path of least resistance, but instead pushing the hard path which is more rewarding and satisfying in the long run - a truly decentralised cryptocurrency which is free from control and corruption.
This is the end goal of bitcoin.org:
http://bittorrent.org/
BitTorrent is what I would call a decentralised piece of software driven by merit. The BitTorrent client marketplace is highly competitive and diverse.
Here's the front page:
Clients page:
If anyone wants to test, then add this to your hosts file:
176.31.24.241 bitcoin.org
Quote
A key concept for the health of a decentralised system is variety. Variety
leads to a richness of characteristics that allow a system to remain
resilient to attack. A good analogy is that of a species, where having a
wild contrast in genetics, doesn't expose that species to any one specific
weakness and limits the effect that a disease can cause. Likewise, Bitcoin
gains an immunity from attack by having contrasting clients that approach
specific problems in different ways.
leads to a richness of characteristics that allow a system to remain
resilient to attack. A good analogy is that of a species, where having a
wild contrast in genetics, doesn't expose that species to any one specific
weakness and limits the effect that a disease can cause. Likewise, Bitcoin
gains an immunity from attack by having contrasting clients that approach
specific problems in different ways.
For all the complaining about centralisation on this forum, people sure seem to be fighting my efforts to promote alternative clients, to avoid promoting a strong exchange competitor by continuing to trade on MtGox and to avoid getting involved with development (bitcoin library with Python bindings is out there). We should not be pushing something because it is simply the path of least resistance, but instead pushing the hard path which is more rewarding and satisfying in the long run - a truly decentralised cryptocurrency which is free from control and corruption.
This is the end goal of bitcoin.org:
http://bittorrent.org/
BitTorrent is what I would call a decentralised piece of software driven by merit. The BitTorrent client marketplace is highly competitive and diverse.
Here's the front page:
Clients page:
Electrum: This client has a focus on being fast, having low resource usage and making it easy to back up your wallet. It runs on Linux and Windows. It's well suited to tech-savvy individuals who want to get started with Bitcoin immediately. It operates in conjunction with a remote server which handles the most complicated parts of the Bitcoin system, which is why it's fast. However, by running this client you don't contribute your computers resources to the core network, and the remote servers that help give it good performance have the ability to see all your transactions and tie them together. Whilst you need provide no personal information to use Electrum (as is true for all Bitcoin apps), this means the privacy level is lower than for other clients such as Bitcoin-Qt. Merchants are recommended to use Bitcoin-Qt or other p2p clients. Electrum is designed for people with a reasonably high level of technical ability.
As the main developer of Electrum, I strongly disagree with this message. It is inaccurate, and it sounds like you are trying to scare users off.
Electrum simplifies the use of Bitcoin, because it removes the need to download the complete blockchain, and to do regular backups of your wallet.
These two aspects of the Satoshi client make it require a substantially higher "level of technical ability" than Electrum.
So, the main question is whether the description of clients should be left to developers or independent reviewers.
FYI, the description I initially provided for the site is the following:
Quote
Electrum is a lightweight client that was designed to simplify the use of Bitcoin. There is no waiting time at startup, because Electrum does not download the Bitcoin blockchain. You do not need to perform regular backups, as your wallet can be recovered from a secret phrase that you can memorize or write on paper. Electrum is available for Linux, Windows and Android.
Thanks Amir and Luke for making this happen. I totally agree this is an important step for the project. AsI said on Lukes pull-req, I prefer the simpler approach so thanks for implementing it. I like the box-based layout, as columns are indeed easier to read and it implies less ordering.
The key part is really the language used. We should try and approach this as a software company would - aggressively eliminate jargon and try and make things as end-user friendly as possible. The current text provided by Amir is better than the grid, but it's still quite developer oriented and discusses details that only programmers care about.
I'm not a developer of any of these clients, except indirectly MultiBit, but here's how I'd rephrase the text. I'm not suggesting my explanations are 100% correct but I think they focus on how end users might perceive them, and it avoids jargon like "frontends" and "blockchain".
There are still some things worth thinking about:
I have tried to present a fair and balanced view of each client. Nevertheless, everyone will want to position their apps in the best possible light. For example, I have described Electrum here as "suitable for people with a reasonably high level of technical ability". The reason is that its website is clearly designed for programmers. The install instructions for the worlds most common OS list how to install Python and its dependencies first, with an actual installer left to some random guy in a forum thread. MultiBit on the other hand has standard installers for all 3 main platforms right on its home page, provided by the core developers. Despite that, as I said, there are still occasional bugs and performance quirks with anything based on bitcoinj (and who knows, maybe other clients too, I haven't used them).
I think we should all be conservative about our own code. BitCoinJ based clients do basically work but there are edge cases where the protocol is not properly implemented, eg, if you pay yourself directly from a coinbase transaction, that currently won't appear (being fixed now). I would be hesitant recommending any client to my mother today because IMHO none of them have the right combination of bulletproof maturity, high performance and dedication to ease of use that would be required. I hope MultiBit will get there first, but I think Electrum could probably also do a good job if it had more of an non-technical user focus.
The key part is really the language used. We should try and approach this as a software company would - aggressively eliminate jargon and try and make things as end-user friendly as possible. The current text provided by Amir is better than the grid, but it's still quite developer oriented and discusses details that only programmers care about.
I'm not a developer of any of these clients, except indirectly MultiBit, but here's how I'd rephrase the text. I'm not suggesting my explanations are 100% correct but I think they focus on how end users might perceive them, and it avoids jargon like "frontends" and "blockchain".
Quote
All clients listed on this page are free to use, open source, audited by the community and none required any registration or personal information to get started. You should feel free to try any or all of them. Each client will have its own wallet and you can send coins between them as you would between two people.
Bitcoin-Qt: The original software written by Satoshi Nakamoto, the projects founder. If you aren't sure which program to pick, this is a good bet. This application is a peer-to-peer client that builds the backbone of the Bitcoin network. It is suited for enthusiasts, merchants, miners, developers and people who want to help support the project. People who run Bitcoin-Qt are first class network citizens and have the highest levels of security, privacy and stability. However, it can be very resource intensive and you should be willing to leave it running in the background so other computers can connect to yours. If your computer is low powered or you aren't willing to tolerate a 24-hour+ initial start time, you should consider other clients. Cutting edge features tend to be implemented in other clients first.
MultiBit: This client supports Windows, MacOS and Linux. Its primary focus is on being fast and easy to use, even for people with no technical knowledge. It has a YouTube channel to help you learn the software, and includes helpful features such as an exchange rate ticker. MultiBit supports many languages such as German, Spanish and Greek. MultiBit synchronizes with the network much faster than Bitcoin-Qt and should be ready for you to use within a few minutes. This is a good choice for non technical users who want an easy to use experience, especially if you use a Mac.
Armory: This client has a focus on advanced wallet management features, such as the ability to construct transactions whilst disconnected from the internet. It operates in conjunction with a Bitcoin-Qt install. It is developed primarily for Linux, but also supports Windows and MacOS. It requires a large amount of RAM to operate and if you use Windows, it requires a 64 bit version. It is a good choice for tech-savvy enthusiasts or merchants who want to try out some of the latest ideas in the Bitcoin world.
Electrum: This client has a focus on being fast, having low resource usage and making it easy to back up your wallet. It runs on Linux and Windows. It's well suited to tech-savvy individuals who want to get started with Bitcoin immediately. It operates in conjunction with a remote server which handles the most complicated parts of the Bitcoin system, which is why it's fast. However, by running this client you don't contribute your computers resources to the core network, and the remote servers that help give it good performance have the ability to see all your transactions and tie them together. Whilst you need provide no personal information to use Electrum (as is true for all Bitcoin apps), this means the privacy level is lower than for other clients such as Bitcoin-Qt. Merchants are recommended to use Bitcoin-Qt or other p2p clients. Electrum is designed for people with a reasonably high level of technical ability.
Bitcoin-Qt: The original software written by Satoshi Nakamoto, the projects founder. If you aren't sure which program to pick, this is a good bet. This application is a peer-to-peer client that builds the backbone of the Bitcoin network. It is suited for enthusiasts, merchants, miners, developers and people who want to help support the project. People who run Bitcoin-Qt are first class network citizens and have the highest levels of security, privacy and stability. However, it can be very resource intensive and you should be willing to leave it running in the background so other computers can connect to yours. If your computer is low powered or you aren't willing to tolerate a 24-hour+ initial start time, you should consider other clients. Cutting edge features tend to be implemented in other clients first.
MultiBit: This client supports Windows, MacOS and Linux. Its primary focus is on being fast and easy to use, even for people with no technical knowledge. It has a YouTube channel to help you learn the software, and includes helpful features such as an exchange rate ticker. MultiBit supports many languages such as German, Spanish and Greek. MultiBit synchronizes with the network much faster than Bitcoin-Qt and should be ready for you to use within a few minutes. This is a good choice for non technical users who want an easy to use experience, especially if you use a Mac.
Armory: This client has a focus on advanced wallet management features, such as the ability to construct transactions whilst disconnected from the internet. It operates in conjunction with a Bitcoin-Qt install. It is developed primarily for Linux, but also supports Windows and MacOS. It requires a large amount of RAM to operate and if you use Windows, it requires a 64 bit version. It is a good choice for tech-savvy enthusiasts or merchants who want to try out some of the latest ideas in the Bitcoin world.
Electrum: This client has a focus on being fast, having low resource usage and making it easy to back up your wallet. It runs on Linux and Windows. It's well suited to tech-savvy individuals who want to get started with Bitcoin immediately. It operates in conjunction with a remote server which handles the most complicated parts of the Bitcoin system, which is why it's fast. However, by running this client you don't contribute your computers resources to the core network, and the remote servers that help give it good performance have the ability to see all your transactions and tie them together. Whilst you need provide no personal information to use Electrum (as is true for all Bitcoin apps), this means the privacy level is lower than for other clients such as Bitcoin-Qt. Merchants are recommended to use Bitcoin-Qt or other p2p clients. Electrum is designed for people with a reasonably high level of technical ability.
There are still some things worth thinking about:
- I agree that for now, we should recommend Bitcoin-Qt as the default option until we have more experience with other clients and their general level of polish and robustness is better. Bitcoin-Qt has an unfortunate name. The UI toolkit used should really not be exposed to end users. I'd support renaming it to something like "Bitcoin Core" or "Bitcoin Classic".
- Not every client supports every OS, so it might make sense to have an OS picker at the start to avoid showing users clients that might sound good, but won't run on their machines. I almost didn't see the tiny icons at the bottom.
- Ordering: above I put MultiBit after Bitcoin-Qt. The reason is that despite its flaws (most of which are in any case my fault), it is the most professional and approachable client, IMHO. It runs on every OS, is translated into other languages, does not have any unexpectedly huge resource requirements and Jim has a sharp focus on the average user. Eg, he prioritized an exchange rate ticker and localization over some of the more interesting wallet features. This does not mean I think other clients are "worse", just that most users who end up at this page will be individuals who want to receive and send a few coins and we should ensure they are catered to first.
I have tried to present a fair and balanced view of each client. Nevertheless, everyone will want to position their apps in the best possible light. For example, I have described Electrum here as "suitable for people with a reasonably high level of technical ability". The reason is that its website is clearly designed for programmers. The install instructions for the worlds most common OS list how to install Python and its dependencies first, with an actual installer left to some random guy in a forum thread. MultiBit on the other hand has standard installers for all 3 main platforms right on its home page, provided by the core developers. Despite that, as I said, there are still occasional bugs and performance quirks with anything based on bitcoinj (and who knows, maybe other clients too, I haven't used them).
I think we should all be conservative about our own code. BitCoinJ based clients do basically work but there are edge cases where the protocol is not properly implemented, eg, if you pay yourself directly from a coinbase transaction, that currently won't appear (being fixed now). I would be hesitant recommending any client to my mother today because IMHO none of them have the right combination of bulletproof maturity, high performance and dedication to ease of use that would be required. I hope MultiBit will get there first, but I think Electrum could probably also do a good job if it had more of an non-technical user focus.
BTW if you want to write a BitcoinMedia about it then give me an email.
Thanks for the offer to contribute to Bitcoinmedia : I will follow up on that asap.
Two questions: is it open-source? Does it store people's keys (wallet) locally on the device?
Our mobile developments are not open sourced until we close our second round of financing: this is a topic we want to keep open with investors we are talking to.
Paytunia mobile is a thin client. The keys are not stored on the mobile device but on the backend (paytunia.com) server.
My experience trying to use embedded keys (stored on a mobile device) taught me that the user experience is better with a true thin client (no keys stored on mobile).
Anyhow, both options (hosted keys or embedded keys) have their pros and cons but none can be said far superior to the other: we will let the market decide.
Bitcoind still has the distinction of being the direct descendant of the client created by the original developers.
It is the same client, just not the same developers.And I assert that it's "reference" because if it's not, then the Bitcoin protocol as a whole is unconscionably underdocumented.
The latter. Case in point: What happens if some other client becomes more popular, and there turns out to be a subtle difference in the implementations that result in a block chain fork?
Then the more popular client wins, and the less popular ones need to correct their implementation to match it.Jump to: