A lot of coins are lost, it should be changed

[khal]

I think you misunderstood. The coins for the expired domains would not go back to the domain's owner. They would go back to the "earth" to be found by miners once again. Of course, I'm using the term "earth" as a mining metaphor.

Oh, this possibility does not come to mind because that means the network must look through the whole blockchain to count lost coins due to registered names and then inject them back. But, number of coins generated per block is a simple algo :
50 coins per block during 4 years, then divide by two and use during 4 years, etc. When to insert lost coins in this scenario ?

A similar thing could be done to guard against lost coins due to lost wallets.

Coins are associated with addresses

Have some means of pinging addresses to check that they are still in use (transactions being made regularly)

If an address has not been used in a certain amount of time, then kill that address and release all coins back into the NMC earth, ready to be mined again.

This has been discussed a lot in bitcoin too. Forcing users to "update" their coins makes no sense. And same problem appears with the current algo.

[drknark]

This has been discussed a lot in bitcoin too. Forcing users to "update" their coins makes no sense. And same problem appears with the current algo.

I have been thinking a bit about this myself. I realize the problem with this. I was thinking maybe something more gradual and very long term than "lose all your coins after x amount of time".

More like, address not active for 5 years -> lose 10% of coins. Another 5 years -> another 10%. Or maybe even 10 years. The point being to look at the long-term health of the system, some coins being lost for 40-50 years is much less severe than being lost forever. That's if we're really looking to make a system that can keep going pretty much forever.

But, I suppose another solution is to implement a finer granularity to the coins in the future, making the atomic unit 10^-16 instead of 10^-8 or something like that.

Anyway, to get back on topic, I definitely think that registration fees should go back into the system

[vinced]

The coins where domains are attached are currently 0.01NMC, but there's nothing in the code that validates that amount. It can be as low as 10^-8. I will change to 10^-8 in the near future.

In the current system, registration fees are reduced by 50% about every 3 months. There is no lower limit, so the fees will go to zero after a few years. If the value of NMC grows faster than that, then indeed there is an issue.

I will consider speeding up the registration fees decrease, probably to 2x per month, to try to match bitcoin/namecoin growth rates. This will be done at the same time that the switch to joint BTC/NMC mining is done.

I apologize it took me a long time to respond. Please expect 1 week response time in the future.

[khal]

In the current system, registration fees are reduced by 50% about every 3 months. There is no lower limit, so the fees will go to zero after a few years. If the value of NMC grows faster than that, then indeed there is an issue.

Oh, i misunderstood a comment in the code ? (namecoin.cpp:701 : // Round up to CENT)
I've updated the formula of http://dot-bit.org/tools/domainCost.php and price seems to vary, due to the modulo 8196 maybe ? Or is it a php problem only ?
Should we remove this rounding too ?

http://dot-bit.org/tools/domainCost.php?block=505120 : 0.00999999
http://dot-bit.org/tools/domainCost.php?block=605120 : 0.06531640

[JohnDoe]

Very nice to know that name_firstupdate doesn't have a lower limit like I thought. Could have sworn I read that somewhere.

[ribuck]

Consequences are :
- all fees goes to miners...

And that's the flaw. It means that large miners (or mining pools) can speculatively register thousands of names "on the cheap" (because they can pay the first_update fee to themselves).

Losing the first_update fee is the perfect way to discourage domain name squatting. Because the first_update fee halves every two months, the number of lost coins won't be a problem.

[JohnDoe]

You are right ribuck, didn't think about that.

Btw, looks like the fee decrease will definitely have to be sped up as NMC value is rising fast. Currently it costs over $20 to register domains (probably much lower if you mine the NMC instead of buying it though).

[vinced]

This is implemented in the latest github commit.

[ribuck]

This is implemented in the latest github commit.

Not happy. It will just lead to massive squatting by miners.

[khal]

Namecoin is designed to provide almost free names at the end. So, this problem is not really a problem but a design choice.
Domains are already reserved by some "squatters", so this will not change a lot of things.

Due to the BTC/USD price and the super fast difficulty increase of namecoin (faster than "planned" by the old algo), a name cost up to 40/60$. Don't you think it's a little bit overpriced and limit it's usage for normal users ?