Bitcoin Forum

==cut out===

Purpose of this thread to find a way to build censorship resistant authoritative key=value storage (Distributed DNS) based on Satoshi ideas.

Please scroll down for design improvement proposals to namecoin.

Also here http://dot-bit.org/forum/viewtopic.php?f=5&t=58&start=40
And russian discussion is here: https://bitcointalk.org/index.php?topic=61486.80

Current namecoin design has several errors which are:

0. Coins are destroyed Proof (http://dot-bit.org/forum/viewtopic.php?f=5&t=58)
1. NetworkFee descreasing for 50 to zero and will be zero in ~80000 block Proof (http://dot-bit.org/forum/viewtopic.php?p=969#p969) and has no feedback with network activity like difficulty has.
2. Namecoin was designed for free domains and this is its future Proof (http://http//dot-bit.org/forum/viewtopic.php?p=763#p763)
3. Developers have to do system intrusion from time to time to avoid collapse or spam hell Proof (http://http//dot-bit.org/forum/viewtopic.php?p=501#p501) Proof (http://dot-bit.org/forum/viewtopic.php?p=334#p334)

If I understood correctly, namecoin inflation never stops. It's a constant, linear growth. So, there will always be new coins.

The greatest problem I see is that this way it seems the domain growth rate is also linear. If one day it gets popular and goes viral, exponential domain growth will not be possible. Namecoin domains will become too expensive, people will give up, and that will slow down adoption to a linear growth as well, what would be a pity.

On the other hand, making domain growth exponential might be too much.

I wonder how to match domain growth with the demand for new domain names...

PS: I don't understand much about namecoin, so I might be saying bullshit.

namecoin has identical emission behavior to bitcoin. 21M of max "coins" (each coin is 1E8 (if not mistake) integer). So the total possible number of namecoins is the same as bitcoins = 21E6 * 1E8 = 21E14. It is deflated in value by default.

Additionally, namecoin has special transactions which contain key=value pair (i.e. domain => ip). When first registering such key (domain), the fee for registering is actually destroyed (to fight squaters). So every domain registered decreases total possible number of NMC currently by 0.01 (1E6 = 1 million integers). 81k namecoins (81E12 integers) are already destroyed by this way.

So namecoin is doomed for death with this behavior someday.

Wow...

If that's the case, frankly, it's awful. How come people didn't think that they would run out of coins when they implemented this?
It seems like one of those "temporary fixes to be improved later", something that should not be present on important protocols...

Actually this is not from empty place. There is some logic in that.

Miner has a 1Ghash/s mining power.

With a difficulty 518525.94130231 he will get 1.94 NMC per day.

Day left, he mined 1.94 NMC and bought 194 domains.

If system pass that 1.94 NMC as reward for block solving, miners' total daily revenue will be increased by 1.94*(his_power/network_power).

Actually this is the "money from nothing". And devs still didnt find any suitable method to resolve this except to destroy these coins.

So my proposal is: may be is it possible to hold domain associated coins util it will be expired.

afaik there's some algorithm which calculates price for domain from last x blocks. So yes, those coins are destroyed, but when namecoin gets some popularity and namecoins will be more expensive, then domain will cost maybe eve 1 satoshi, which means we can still have bazilion of domains.

I am programmist (not in c++ unfortunately), and one of programmist rule says: The future comes much quicker than expected.

So if the future will bring a collapse, im worried. Worried because namecoin may become a basis of I2P naming structure.

I see, still we're talking about almost 21.000.000.000.000.000 domain updates in current system.

However I agree that never ending inflation may work better than hardcoded 21mil limit. Namecoin isn't a currency and I don't see real reason to limit coins in existence (especially when system is systematically destroying them).

This is also a point!

Regardless if we have a real problem or not, the possible problem can't be much of an issue right now, right?

A few days ago I bought 1000 Namecoins for about 4.6 Bitcoins, i.e. 30 USD or something. For this amount I can register thousands of domains (about 5000 if I remember correctly). So, stopping squatters clearly doesn't work, right? Perhaps we should INCREASE the destruction of namecoins, if squatters are the real issue here?

Actually khal's explanations confused me. He said the problem in squaters, but the real problem is in money from nothing. That is why they destroy coins.

Squaters are not a problem.

The problem that system by current design is DEAD in a future. And my programmer expirence says that future becomes much quickly than expected.

After 500k blocks fee will be 0 if I don't mistake. What about squaters? ))


People ask me, what do I dont like.

This all smells like the central person will decide what fee will be, because it doesn't have its algo, I don't see any documents explaining effects of this code in future.

What will be in a future? What is the diagram of number of lost namecoins versus number of mined? Where is the intersection?

Very, very many questions.

Ok. I think I got it.

If system will pass netfee to miners, it will inflate to spamhell
If system will have a zero netfee, it will be a spamhell

So it is doomed for controlled self-killing by destroying netfee coins, and amount of this netfee is assigned by some central person in patches. He have to monitor the network and keep netfee in reasonable range. Like federal reserve with its Discount rate.

In sysadmins sociaty this method is called "spike-nail".

Good night.

I've parsed a block chain to know the dynamic grow of registered domains and dynamic grow of lost coins.

I tried to look in the future by mathcad predict() function. And got a lulz :) :) :)

Left graph - number of registered domains
Right graph - number of lost coins

Red line - current data
Dot line - mathcad's prediction :)

(this not the correct prediction, just lulz)
http://img862.imageshack.us/img862/7093/mathcad.jpg (http://imageshack.us/photo/my-images/862/mathcad.jpg/)

Anyway,

GetNetworkFee() is a function which taken from the sky, and returns a current network fee which is guessed by developers. GetNetworkFee() -> 0 @ block 80000
first_update destroys GetNetworkFee() coins, increasing deflation

when GetNetworkFee will be zero, namecoin will be full of spam. To avoid it, developers should watch and update this fee with relevant values and ask all network members to update their clients.

developers is the central point of failure in this system. The system is controlled by developers intrusion by changing network fee for name registration, like US economics controlled by Federal Reserve intrusion by changing discount rate.

Also i noticed that this question is uncomfortable for developers. They don't try to go deep into problem, but giving common answers.

The problem with destroyed coins exists from beginning and still not resolved and ignored by developers.

NameCoin is a system with unpredicted behavior. It is not suitable for usage in i2p with a current design.

I am sorry.

There is only one developer and as far as I know they haven't responded to you yet - I don't know if that's an indication of being uncomfortable or not. You've been appealing to 'khal', but the namecoin developer is vinced (https://bitcointalk.org/index.php?action=profile;u=9552). Have you looked through the namecoin thread (http://bitcointalk.org/index.php?topic=6017.0) to see if there's any reasoning?

Well thank you (not) for spewing your learning curve of namecoin all over the forum ... unsuprisingly, from following your chaotic thought train, you've arrived at  the wrong conclusion.

namecoin is doing fine.

bitcoin also has a central core of developers (actually mostly just gavin) deciding on minimum transaction fee ... gonna go dump all over on that now also??

Bitcoin developers does not decide how much coins to destroy, bitcoin doesn't destroy coins at all. Am talking about network fee, not transaction fee.

Also I am glad that namcoin is fine, but still no answer.

Network fee falling to zero and what you gonna do?

Should i put a client on local machine with gigabytes of free domains for sale?

Okay. Let my chaotic mind continue to travel from critic station to the solution proposal station.

How bitcoin deals with increasing bounty hunters? It has a positive feedback between network power and work difficulty. I.e. number of people (actually their cummulative power) grows => difficulty grows. Lets see

http://img543.imageshack.us/img543/2343/mbitcoin.png (http://imageshack.us/photo/my-images/543/mbitcoin.png/)

And what is namecoin in current stage? It is a addition to bitcoin, which charges fees and issues a TLDs. But where to put that charged fees? If we return them to miners, then there will be parasite positive feedback between number of ordered TLDs and bounty increasement. It will let miners to order TLDs only with purpose of their bounty increase. Free Money. Like beer.

So devs decided to destroy that fee. If we destroy that fee, the miners (and all others) will become bit reacher also. But this will also increase fees in their value. So this is some sort of positive feedback between number of ordered TLDs and fee increase.

But with bad implementation and negative effects:
1) System changes fees descrease according to dummy algo without feedback of its growing
2) The 1) effects are that devs should regulate this algo by hands doing network intrusion
3) If they will not properly guess fees, system will be full of spam or collapse
4) By destroying fees, system destroys itself and forces NMC value to have unpredicted behavior

http://img52.imageshack.us/img52/871/mnamecoin.png (http://imageshack.us/photo/my-images/52/mnamecoin.png/)

The proposal
1) Registration fee must be constant (this needs some mathematics check)
2) Registration fee should return to turnover as miners' bounty
3) To avoid described parasite feedback, we need to put another feedback between number of TLDs registered for a certain time period and difficulty.

So if the registration activity will grow, the bounty will grow, but difficulty will also grow.
Benefits:
1) System do not destroy coins
2) devs do not do system intrusion
3) System regulate itself

http://img844.imageshack.us/img844/5590/mnamecoin2.png (http://imageshack.us/photo/my-images/844/mnamecoin2.png/)

In this figure, namecoin becomes even better than bitcoin. As it is also self-regulated and coins can be directly converted to "Internet Real-Estate" - domains. Just like gold.

Unlike bitcoin, we can cancel all fees for regular transactions at certain age. When all coins will be mined and transactions will be canceled, the miners will continue work as the domain registration never ends (domain records have a TTL). And domain registration and updates supplies miners with coins. And miners supplies domains with security.

And this gold mining will never ever end. Amen.

interesting  :)
i always wondered what namecoin would do after all the coins where mined, but supposedly the fees would increase and first update would cost you zero, so nmc coins are only destroyed in it's first years of life

But not ideal. Vulnerable to name registration DoS, which can spike difficulty increase - as suggest in IRC. This figure missed something and needs a brainstorm.

We have a couple of problems on previous fig:

1. From where to get a domain transaction fee? And where to link it?
2. Coins for domain transaction must be paid to miners back for their efforts, but tricky to avoid inflation.
3. Nedd an ability to accept even million of domain transactions per block
4. We must not intrude to standard N blocks/hour rule. This is the fundamental thing.

I draw a figure after reading BitDNS thread (https://bitcointalk.org/index.php?topic=1790.msg28696#msg28696).

===cut===

Doesn't work. Main bounty price is decreased, so domain price also collapsed to zero. Need improvement

I'm quite devastated by this topic.

It seems that namecoin is doomed from the start. Why didn't the NMC developers account for that ?

This concept design (https://bitcointalk.org/index.php?topic=64279.0) demonstrates this figure and not vulnerable to registration ddos

Last time I checked, block diagrams do not provide mathematical proofs, just abstract schematics ....

If you good in mathematics, you are welcome to check structure. Some negation calcs were posted in russian thread, but we dealed with them.

This! Allowing perhaps millions of transactions per block would open up the possibility of something being done with m/ (messages) namespace. Pruning them seems like the ideal solution if it's an instant messaging scenario, but keeping them around would be better if it's an email sort of scenario. Either way you'd need lots of transactions to be possible.

Imagine a decentralized  IRC or IM client where you control your security with encryption and no centralized authority censors,limits access,or monitors messages that are difficult to encrypt without plugins.

For example, say you want to have a secure place to discuss terms, perhaps you need to give some personal phone or address information, or website access passwords, and don't want it getting into the wrong hands. You can use their n/ name to verify who they are with a GPG key and/or signed message verifying their ownership of a public bitcoin address, you can then discuss terms with a secure namecoin instant messaging client (encrypted text that if desired gets pruned from blockchain), then you send your bitcoins to the address you verified belongs to the person you're talking to.

If problems arise, the agreement is stored in the block chain, and you can provide the decryption key to anyone to verify that the other party has not fulfilled their obligations.

I understand that i left some security holes in there but you get the general idea. I think it would be very cool. :)

/end lofty dreams for namecoin

Some limits are so large they are not a problem ever.

Some day we may exhaust the IPv6 space... Still the internet happily exists on IPv4 mostly.

Some day even all conventional DNS top level domains may be taken. Proof: there are only limited characters allowed (c) and there's some limit on segement length (l), so there are only c**n different *.com domain names (repeat for others). If I estimate (from memory) c=30, l=63, that's roughly 1E93 domains, and most of them are pretty much unusable as memorizable names. Are you crying doom on DNS too, because the future may arrive incredibly early?

If anything, I'd like more namecoins destroyed for upkeep instead just initial registration to mitigate squatting.

-coinft.

How much namecoins domain costs now? And how much namecoin costs itself?

And what will be in 6 months?

suggestion: free domains and spam hell.