Namecoin Flaws & Improvements Tip Bounty

[bytemaster]

Namecoin has been around for a long time and from what I can tell has failed to gain as much traction as the concept deserves.   I am looking to get feedback on how to develop a far better system than namecoin and want to establish the criteria for this system.   First I would like to suggest some of the flaws I see with namecoin and get feedback on some of my proposed solutions:

1) Fixed Price Names -  the value of a name is similar to real estate and any name registration system that attempts to fix prices will either set them too high (no one registers) or set them too low (domain squatting).   

2) Too Slow or Too Fast.....  the rate at which names can be registered is either too slow for the demand or too fast and thus squatters.

3) Too hard to use  - complete lack of any GUI

Proposed Solutions:
------------------------
1)  Start with a crypto-currency that is mined into existence.
2)  New names are all auctioned off to the highest bidder.  Only after 72 hours with no bids is the name considered owned.    This will prevent squatting because speculators would have to pay more for 'good names' than 'throw away names' due to the bidding process.   This will also ensure names are put to the highest use.
3)  Fees from the initial bid are split between the the miner and dividends paid to those who hold the crypto-currency
4)  If someone outbids the initial bid, then half of the 'profit' is given to the original bid and half paid to the miner and dividends.
5)  Allow people to put their names up for auction or list them for sale.

Combine this with an easy to use interactive GUI to show trending names and you have a system where many different users can speculate on names playing an addictive game of 'price is right' attempting to bid the highest without going over what someone is willing to pay for a name and profit while they are at it.  If they over-bid then they can always sell at a loss to the next person in line after the auction is over.

Economic Consequences?
-----------------------------
No one has incentive to 'squat' on names because they are better off owning the crypto-currency and receiving dividends from all of the names that are actually purchased rather than speculate on any individual name.   Owning this crypto-currency is like having a stake in all unclaimed names.

Everyone has incentive to 'bid up' high value names either to profit from reselling them and to increase dividends paid. 

Names can be registered at any rate based upon supply and demand.   

No-value names like bytemaster1234 would not be bid up, while high-value names like apple, google, and bitcoin would be bid up. 

Here is what I am looking for and will pay tips for:

1) New criteria / requirements that a decentralized name registration system must meet.
2) Significant design flaws with Namecoin that should not be repeated.
3) Useful and Constructive Feedback on my proposed solution.   
4) Reasons why you haven't used Namecoin and what it would take to convince you to use my system.

Just a reminder that I have a solid track record for actually paying tips & bounties and if you are not satisfied with allowing me to be the sole arbiter on to whom and how much to pay then please do not participate in this thread.

[1715029390]

1715029390

[snailbrain]

Namecoin has been around for a long time and from what I can tell has failed to gain as much traction as the concept deserves.   I am looking to get feedback on how to develop a far better system than namecoin and want to establish the criteria for this system.   First I would like to suggest some of the flaws I see with namecoin and get feedback on some of my proposed solutions:

1) Fixed Price Names -  the value of a name is similar to real estate and any name registration system that attempts to fix prices will either set them too high (no one registers) or set them too low (domain squatting).  

2) Too Slow or Too Fast.....  the rate at which names can be registered is either too slow for the demand or too fast and thus squatters.

3) Too hard to use  - complete lack of any GUI

Proposed Solutions:
------------------------
1)  Start with a crypto-currency that is mined into existence.
2)  New names are all auctioned off to the highest bidder.  Only after 72 hours with no bids is the name considered owned.    This will prevent squatting because speculators would have to pay more for 'good names' than 'throw away names' due to the bidding process.   This will also ensure names are put to the highest use.
3)  Fees from the initial bid are split between the the miner and dividends paid to those who hold the crypto-currency
4)  If someone outbids the initial bid, then half of the 'profit' is given to the original bid and half paid to the miner and dividends.
5)  Allow people to put their names up for auction or list them for sale.

Combine this with an easy to use interactive GUI to show trending names and you have a system where many different users can speculate on names playing an addictive game of 'price is right' attempting to bid the highest without going over what someone is willing to pay for a name and profit while they are at it.  If they over-bid then they can always sell at a loss to the next person in line after the auction is over.

Economic Consequences?
-----------------------------
No one has incentive to 'squat' on names because they are better off owning the crypto-currency and receiving dividends from all of the names that are actually purchased rather than speculate on any individual name.   Owning this crypto-currency is like having a stake in all unclaimed names.

Everyone has incentive to 'bid up' high value names either to profit from reselling them and to increase dividends paid.  

Names can be registered at any rate based upon supply and demand.  

No-value names like bytemaster1234 would not be bid up, while high-value names like apple, google, and bitcoin would be bid up.  

Here is what I am looking for and will pay tips for:

1) New criteria / requirements that a decentralized name registration system must meet.
2) Significant design flaws with Namecoin that should not be repeated.
3) Useful and Constructive Feedback on my proposed solution.  
4) Reasons why you haven't used Namecoin and what it would take to convince you to use my system.

Just a reminder that I have a solid track record for actually paying tips & bounties and if you are not satisfied with allowing me to be the sole arbiter on to whom and how much to pay then please do not participate in this thread.

namecoin is only just starting to "pickup".. the amount of working going into it this last 3 months is amazing. http://dot-bit.org/forum/viewforum.php?f=2

do you read the dot-bit forum??

Currently there is a Qt with all the "goodies" integrated..
The code is being updated to the latest bitcoin.. http://dot-bit.org/forum/viewtopic.php?f=5&t=1009
all fees are being discussed: http://dot-bit.org/forum/viewtopic.php?f=5&t=1016
There will be some hard fork needed...

why don't you contribute to namecoin?

anyway..

1. the price of domains was already set Very High to prevent (slow down) squatting.. to purchase a name was 50 NMC.. people had their chance? ... also.. you could just create another tld within namecoin.. .. also with the new fees, renews will be more expensive so this will slow down people with 1000s of names.. hopefully. or you can just suggest something on the dot-bit forum.

2. Registering names is ok i think (speed wise)..

3. Namecoin-qt .. this is the current GUI with intergrated name registration https://bitcointalk.org/index.php?topic=236375.0

here you can see amount of names registered recently http://nx.bit

edit: not saying your system isn't interesting

[bytemaster]

When I refer to the speed of registering names I am really referring to the following:

1) Assuming a fixed bandwidth limit that keeps the system decentralized, the system is limited to a couple hundred names per block.   

2) Given the original release pace name registration limit was restricted by the money supply.

Imagine that BitMessage became as usable a regular email and people wanted to sign up at a rate of 100K per day or more, the Namecoin could not handle that load.

Here is what I am looking for and will pay tips for:
1) New criteria / requirements that a decentralized name registration system must meet.
2) Significant design flaws with Namecoin that should not be repeated.
3) Useful and Constructive Feedback on my proposed solution.   
4) Reasons why you haven't used Namecoin and what it would take to convince you to use my system.

I appreciate your post, but it didn't provide any of the things I was looking for.

[snailbrain]

When I refer to the speed of registering names I am really referring to the following:

1) Assuming a fixed bandwidth limit that keeps the system decentralized, the system is limited to a couple hundred names per block.  

2) Given the original release pace name registration limit was restricted by the money supply.

Imagine that BitMessage became as usable a regular email and people wanted to sign up at a rate of 100K per day or more, the Namecoin could not handle that load.

Here is what I am looking for and will pay tips for:
1) New criteria / requirements that a decentralized name registration system must meet.
2) Significant design flaws with Namecoin that should not be repeated.
3) Useful and Constructive Feedback on my proposed solution.  
4) Reasons why you haven't used Namecoin and what it would take to convince you to use my system.

I appreciate your post, but it didn't provide any of the things I was looking for.

no problem.. sorry if i sounded negative...

I posted because it seemed you had not been following Namecoin recently, as your post suggested the project had sort of died and no one was using it.. yet it's actually improving immensely, many people are only just starting to realize it is not "JUST" an altcoin.

signing up at a rate of 100k a day would be more popular than facebook? there are not even 100k namecoin names registered atm. A Gui has only been available for a month.

I believe the current system is fair (some fees need revising).. albeit people don't like the squatters.. having an auction system will only benefit the rich?

4) There hasn't been a real need for a decentralized naming system until recently, this is probably why most people probably haven't used it.. and also i believe one of the issues was lack of a GUI and software to utilize it to it's full potential (of which work is ongoing daily)

just in the last month or 2 there has been drastic improvements (additions?).

examples:
http://veclabs.fuzziqersoftware.com/files/convergence/Namecoin_Poster_2.pdf
http://dot-bit.org/forum/viewtopic.php?f=2&t=1013
http://dot-bit.org/forum/viewtopic.php?f=8&t=525&start=20

[bytemaster]

just in the last month or 2 there has been drastic improvements.

examples:
http://veclabs.fuzziqersoftware.com/files/convergence/Namecoin_Poster_2.pdf
http://dot-bit.org/forum/viewtopic.php?f=2&t=1013

That poster was very well done and I like what they are doing with that convergence FF extension.  

I saw some place that iTunes had 500K new accounts being created per day, perhaps that was just after launch.   All I know is that there are more email addresses than people in this world and to simply give every adult in the US a name within the next 3 years would require very rapid name registration and the required bandwidth would easily consume most low-end DSL lines, particularly for people that live 30 minutes from town.   I mention this because I want a name system to scale to world-wide adoption without overloading a 1 MBit connection with Namecoin traffic alone.  

The other problem with Namecoin is that like bitcoin, the transaction history has to be kept forever without limit.   Perhaps you will need to implement pruning.

Lets just say I really like the idea of a block-chain based name registration system, but am not sure the changes being proposed / discussed in the namecoin forums are sufficient to handle the ultimate demand.

[snailbrain]

just in the last month or 2 there has been drastic improvements.

examples:
http://veclabs.fuzziqersoftware.com/files/convergence/Namecoin_Poster_2.pdf
http://dot-bit.org/forum/viewtopic.php?f=2&t=1013

That poster was very well done and I like what they are doing with that convergence FF extension.  

I saw some place that iTunes had 500K new accounts being created per day, perhaps that was just after launch.   All I know is that there are more email addresses than people in this world and to simply give every adult in the US a name within the next 3 years would require very rapid name registration and the required bandwidth would easily consume most low-end DSL lines, particularly for people that live 30 minutes from town.   I mention this because I want a name system to scale to world-wide adoption without overloading a 1 MBit connection with Namecoin traffic alone.  

The other problem with Namecoin is that like bitcoin, the transaction history has to be kept forever without limit.   Perhaps you will need to implement pruning.

Lets just say I really like the idea of a block-chain based name registration system, but am not sure the changes being proposed / discussed in the namecoin forums are sufficient to handle the ultimate demand.

Pruning will be implemented (is planned).. after Khal rebases the current code on the latest bitcoins.

Expired names will be pruned, and also possibly names with variable expiration times (some short some long).

I think by the time Namecoin comes of age, people will have more than 1mb connections.. my mobile phone has 4g... (ok. it doesn't, but i could have it lol).

[bytemaster]

just in the last month or 2 there has been drastic improvements.

examples:
http://veclabs.fuzziqersoftware.com/files/convergence/Namecoin_Poster_2.pdf
http://dot-bit.org/forum/viewtopic.php?f=2&t=1013

That poster was very well done and I like what they are doing with that convergence FF extension.  

I saw some place that iTunes had 500K new accounts being created per day, perhaps that was just after launch.   All I know is that there are more email addresses than people in this world and to simply give every adult in the US a name within the next 3 years would require very rapid name registration and the required bandwidth would easily consume most low-end DSL lines, particularly for people that live 30 minutes from town.   I mention this because I want a name system to scale to world-wide adoption without overloading a 1 MBit connection with Namecoin traffic alone.  

The other problem with Namecoin is that like bitcoin, the transaction history has to be kept forever without limit.   Perhaps you will need to implement pruning.

Lets just say I really like the idea of a block-chain based name registration system, but am not sure the changes being proposed / discussed in the namecoin forums are sufficient to handle the ultimate demand.

Pruning will be implemented (is planned).. after Khal rebases the current code on the latest bitcoins.

Expired names will be pruned, and also possibly names with variable expiration times (some short some long).

I think by the time Namecoin comes of age, people will have more than 1mb connections.. my mobile phone has 4g... (ok. it doesn't, but i could have it lol).

Even if everyone had a least 2 mb upload, I doubt many would want to run software that consumed 50% of it.   What is your timeline for Namecoin going mainstream? 

[snailbrain]

just in the last month or 2 there has been drastic improvements.

examples:
http://veclabs.fuzziqersoftware.com/files/convergence/Namecoin_Poster_2.pdf
http://dot-bit.org/forum/viewtopic.php?f=2&t=1013

That poster was very well done and I like what they are doing with that convergence FF extension.  

I saw some place that iTunes had 500K new accounts being created per day, perhaps that was just after launch.   All I know is that there are more email addresses than people in this world and to simply give every adult in the US a name within the next 3 years would require very rapid name registration and the required bandwidth would easily consume most low-end DSL lines, particularly for people that live 30 minutes from town.   I mention this because I want a name system to scale to world-wide adoption without overloading a 1 MBit connection with Namecoin traffic alone.  

The other problem with Namecoin is that like bitcoin, the transaction history has to be kept forever without limit.   Perhaps you will need to implement pruning.

Lets just say I really like the idea of a block-chain based name registration system, but am not sure the changes being proposed / discussed in the namecoin forums are sufficient to handle the ultimate demand.

Pruning will be implemented (is planned).. after Khal rebases the current code on the latest bitcoins.

Expired names will be pruned, and also possibly names with variable expiration times (some short some long).

I think by the time Namecoin comes of age, people will have more than 1mb connections.. my mobile phone has 4g... (ok. it doesn't, but i could have it lol).

Even if everyone had a least 2 mb upload, I doubt many would want to run software that consumed 50% of it.   What is your timeline for Namecoin going mainstream?  

not sure i understand why everyone using it would need 2mb upload? (edit: 1mb)

how longs a piece of string

[bytemaster]

If you are on a P2P network and sending one byte for every byte you receive, then you would need that kind of upload capacity.

Now there is always the option of light clients.

[marcus_of_augustus]

Point 3) is wrong. Namecoin has a GUI Namecoin-qt ... (not sure how you missed that one).

There are at least 2 or 3 other threads that post about "namecoins flaws" etc ... perhaps find those to see what happened to the others who tried to redesign/reimplement namecoin from the beginning and/or if your ideas are at all original. If you find anything useful there you can keep your "tip bounty".

Or you could also submit an issue to the namecoin github page if this is a genuine critique to help/contribute to the project https://github.com/namecoin/namecoin/issues

[gmaxwell]

Uh. How about: Not possible to build a secure (zero trust) lite resolver.  To safely and correctly resolve namecoin names you need to run a full namecoin node.

It's solvable but it needs at least a soft-forking block validity rule to fix.

I described the starting point of how to solve this years ago which has later evolved into the various committed utxo proposals, but not a whole lot of actual code.

[bytemaster]

Uh. How about: Not possible to build a secure (zero trust) lite resolver.  To safely and correctly resolve namecoin names you need to run a full namecoin node.

It's solvable but it needs at least a soft-forking block validity rule to fix.

I described the starting point of how to solve this years ago which has later evolved into the various committed utxo proposals, but not a whole lot of actual code.

This is a significant problem that I am aware of, but did not mention above so will send a tip to your donation address. 

I have two proposed 'solutions':
1) restrict the bandwidth and disk space available to something that would allow anyone to run a full node.
2) query many nodes and always pick the most recent transaction.

A merkle tree of all the last trx for each name with with 100 M names would be 4 GB.  And calculating it from scratch would be very CPU intensive.  I suppose this could be viable for heavy-weight servers to maintain.

[domob]

As also snailbrain has already said, I don't think auctions are a good way to distribute names.  This would just benefit large NMC holders (early adopters or those who are rich enough to buy loads of coins later on), who would then be in a position to control what names the community gets.  Even if you mentioned that user names like bytemaster or domob wouldn't be of much value, I presume to you and me they are.  So what would you think about someone listening to the network and out-bidding you on bytemaster just because he/she has lots of coins?  They may even intend to sell the name to you after the fact, because your initial registration request proves to them you value the name and would possibly pay some extra amount to actually get it.

As far as I understand it, this is actually also the reason why in namecoin you have to issue two transactions to register a name - a first one with just a hash and after some confirmations the actual claim.  This prevents nodes or even miners to snatch away names for which they just saw a (still unconfirmed) registration transaction.  I honestly don't think this would be a fairer system - but I admit that the current registration cost could/should be raised to prevent squatting.  Regarding scalability, I think this is not something to worry about right now - as mentioned already in posts above (and by yourself) there are ways to address that which can be implemented in the future when the issue becomes pressing, but for the current user base it should be fine as it is right now.  Development is ongoing, and when the code will be ported to the bitcoin 0.8 base (which is IMHO more important at the moment) there may be time to work on better scalability.

[phelix]

3)  Fees from the initial bid are split between the the miner and dividends paid to those who hold the crypto-currency

Hmm, this would cause weird incentives for miners and stakeholders to pass through discounted name registration. The easiest solution for this is to destroy the fees. As Satoshi put it: "Think of it as a donation to everybody"

4)  If someone outbids the initial bid, then half of the 'profit' is given to the original bid and half paid to the miner and dividends.

Very interesting idea. If somebody outbids you, you at least get a compensation. Also bidding is costly as long as the issue above does not occur.

Please note this is much more complicated than Namecoin is already now. With merge mining and name registration Namecoin is probably the most complex coin already now (As I understood Khal there is more difference between Bitcoin and Namecoin than between Bitcoin v0.3 and v0..

Uh. How about: Not possible to build a secure (zero trust) lite resolver.  To safely and correctly resolve namecoin names you need to run a full namecoin node.

It's solvable but it needs at least a soft-forking block validity rule to fix.

I described the starting point of how to solve this years ago which has later evolved into the various committed utxo proposals, but not a whole lot of actual code.

What about a semi SPV client that would download all full blocks but throw away everything but name info? It could only accept domain name transactions as valid once they are buried beneath 10 blocks or so not matter if there are later changes. With DNS it is rarely a problem if your data takes some time to propagate - it should still be faster than ICANN propagation.

Also it is a big improvement IMHO if you only need to trust your provider's Namecoin server (just like his mail server) instead of centrally controlled authoritay.

[bytemaster]

3)  Fees from the initial bid are split between the the miner and dividends paid to those who hold the crypto-currency

Hmm, this would cause weird incentives for miners and stakeholders to pass through discounted name registration. The easiest solution for this is to destroy the fees. As Satoshi put it: "Think of it as a donation to everybody"

It seems like stakeholders and miners would want the highest possible initial bid, not discounted bids?  If I see two registrations for a name I will pick the highest.  Note that destruction of the money is effectively a dividend as it reduces the supply.  In the end this could result in a large portion of the money supply being destroyed.

[snailbrain]

https://bitcointalk.org/index.php?topic=1790.222

maybe here you will find the original information of the birth of namecoin and alternatives

[phelix]

3)  Fees from the initial bid are split between the the miner and dividends paid to those who hold the crypto-currency

Hmm, this would cause weird incentives for miners and stakeholders to pass through discounted name registration. The easiest solution for this is to destroy the fees. As Satoshi put it: "Think of it as a donation to everybody"

It seems like stakeholders and miners would want the highest possible initial bid, not discounted bids?  If I see two registrations for a name I will pick the highest.  Note that destruction of the money is effectively a dividend as it reduces the supply.  In the end this could result in a large portion of the money supply being destroyed.

Imagine I am a miner with a strong 5% network hashrate. Instead of making a public bid of 1BTC for a name you ask me in private. If I mine your transaction I get 50% of it back (minus ~2% orphaned block rate). I can easily give you a discount of 25% and still make 25% profit.

It might be possible to delay the fee payout to miners of on or more later blocks but once again it makes things more complicated. It may not always be easy to spot hidden incentives of such a system.

[bytemaster]

As also snailbrain has already said, I don't think auctions are a good way to distribute names.  This would just benefit large NMC holders (early adopters or those who are rich enough to buy loads of coins later on), who would then be in a position to control what names the community gets.  Even if you mentioned that user names like bytemaster or domob wouldn't be of much value, I presume to you and me they are.  So what would you think about someone listening to the network and out-bidding you on bytemaster just because he/she has lots of coins?  They may even intend to sell the name to you after the fact, because your initial registration request proves to them you value the name and would possibly pay some extra amount to actually get it.

There are two different issues that I will address separately... early adopters take a huge risk by mining / promoting / investing in a new crypto-system.   It would be unfair if late-adopters were on the same footing as early adopters.   Note, even with name coin early adopters get to squat the good names relatively cheaply because they can buy up the namecoin.  So lets take 'fairness' out of the equation and focus on highest economic utility which actually benefits everyone.  At any time you can buy into the network and start receiving dividends from every name sold.   The longer you hold your coin the more purchasing power you will have.  

So lets see if it makes economic sense to outbid me in the purchase of 'bytemaster'.  Lets assume that it is worth $100 to me, but near 0 to everyone else, so I enter an initial bid of $1.0   Because you know it is worth *something* above 0 to me and you see that I have been using this handle and built up some reputation on it you bid $50.  I just made $24 dollars from you outbidding me.   So I place a new bid at $55 and you receive $52.50.   You still think I will pay more so you bid $75 and I receive $65 dollars.  I finally decide to bid $100 and you get  $87.50.    

At this point I have paid no more than what the name was worth *to me* so I have not been cheated.  Lets look at what everyone made/paid:
Me: -1 + $25.5 - 55 + 65 - 100                     => -$65.5
Network: $1 + $24.5 + $2.50 + $10 + $12.50 => $50.5
You:  -50 + 52.5 - 75 + 87.5                        => $15.00

Now lets assume you are some rich early adopter so you decided to bid it up even higher to say $110.
Now I have made $39.50 and you have paid $95 while the network made $55.50.  
Recognizing that I have a $39.50 profit, I can safely bid $139.50 and still walk away for $100 cost and you would have made $44.50.

So you decide to call my bluff and bid  $150, so I end up making $44.50, you get the name, and the network made $61.00.   You had to pay $105.50 net out of pocket.  I will not out bid you because the most I can afford to bid is $144.50 without my costs exceeding the $100 I was willing to pay for my name.

So the question becomes:  now that you own "bytemaster" for $105.50 at a net cost to you of $105.50...  what are you going to do with it?    You could go around pretending to be 'me', but we are assuming you are just some rich early adopter who is trying to profit off of late adopters and control who gets what name and how much they pay.    Now you end up with a name that only has value to me.  I if you don't turn around and sell it to me at a price I am willing to pay, I will take my $44.50 profits and attempt to buy bitemaster instead.   Once I own bitemaster I will no longer be in the market for bytemaster and therefore, the market value of 'bytemaster' will have fallen dramatically, perhaps 75% or more.  

Conclusion:  Everyone should be able to buy the name they want for a price they are willing to pay and anyone attempting to bid up a name too high purely for speculation purposes, would end up taking a loss.  

There are two ways to allocate scarce resources:  market or ration/price controls.    Attempts to fix the price too low will cause all of the names to be bought up by speculators who will then create an secondary market for the names.   If you fix the price too high, then people will simply not buy as many names as the market could otherwise use.    So what I conclude is that you *always* have an auction for names even with namecoin.   The problem is that namecoin price fixes all names to the same price and only 'rich speculators' can afford to buy enough names to profit 'on average'.     With my system, everyone can buy a share in the 'name squatting game' simply by holding the currency.  So I believe what I propose is inherently more fair.








 

[bytemaster]

3)  Fees from the initial bid are split between the the miner and dividends paid to those who hold the crypto-currency

Hmm, this would cause weird incentives for miners and stakeholders to pass through discounted name registration. The easiest solution for this is to destroy the fees. As Satoshi put it: "Think of it as a donation to everybody"

It seems like stakeholders and miners would want the highest possible initial bid, not discounted bids?  If I see two registrations for a name I will pick the highest.  Note that destruction of the money is effectively a dividend as it reduces the supply.  In the end this could result in a large portion of the money supply being destroyed.

Imagine I am a miner with a strong 5% network hashrate. Instead of making a public bid of 1BTC for a name you ask me in private. If I mine your transaction I get 50% of it back (minus ~2% orphaned block rate). I can easily give you a discount of 25% and still make 25% profit.

It might be possible to delay the fee payout to miners of on or more later blocks but once again it makes things more complicated. It may not always be easy to spot hidden incentives of such a system.

What good does this do you?   Just because you get your transaction in doesn't mean you own the name yet.  I could come along after you any time in the next 1000 blocks (72 hours) and out bid you with a new transaction in the blockchain.

[snailbrain]

@bytemaster

I thought you said people didn't want to pay for email accounts? (assuming bitmessage used your naming system).

also.. if you want people to buy names faster than you can with namecoin, surely an auction system is actually slower?

also.. people can auction namecoin names if they wanted.. all we need is something like an exchange for names.. i was thinking of setting a site up to do this recently