Registering is too cheap

[phelix]

As expected the namecoin price went to the bottom because of merged mining. Thanks for devaluating my coins btw.

As of the domain cost calculator registering a domain costs only 1.32303823 NMC = 0.02BTC = 0.04USD at the moment. This is much too cheap, people will be buying up all names available and bring the whole project to a painful grinding halt.

I think the price should be at least 50 NMCs again.

[khal]

And if nmc value increase, we should decrease the registration cost again ? And vice-versa ?

We may find a way to adjust it depending on some non-external criteria (ie: nmc price in btc is an external value not know by the network).

Edward_Black proposed some method to me, i'll point him to this thread. It may be interesting to analyze it and to search for other methods.

But, this problem may be unsolvable... because we want for names to cost as less as possible to allow millions of registrations. And, registration fees were here to avoid a rush at the beginning, don't forget that.
We must also define what is "registration spam" and what is not. Where is the limit ? If we are too strict, this will limit namecoin usage. If we are to low, a lot of domains will be reserved to be sold later...

Aims :
- allow millions of registrations (if there is no namecoin left [extreme example], this is a problem)
- avoid very low registration fees which allow registration spam

Limits :
- do not depend on external parameters
- namecoin doesn't know the value of 1 NMC in BTC

[moa]

The market appears to be working, (it did not go to zero). Now is not a good time to start interfering again. It would risk removing all credibility from namecoin forever.

We will always get people who think it is too cheap or too dear. If you think it is too cheap, buy, if you think it is too dear, sell, simples. Stop whinging. Should we listen to every complaint about market price or should we listen to none?

[doublec]

IIRC the design docs mention it is the transaction fees that should control the price of the domains once the network fee dropps low. So big miners (probably the large merged mining pools) can increase the cost of domains by only accepting registrations with a large transaction fee. Those that want fast registration can pay the fee. Those willing to wait for a smaller fee miner to pick it up can pay the standard fee. So if you want the price to increase, lobby the big miners.

[phelix]

The market appears to be working, (it did not go to zero). Now is not a good time to start interfering again. It would risk removing all credibility from namecoin forever.

We will always get people who think it is too cheap or too dear. If you think it is too cheap, buy, if you think it is too dear, sell, simples. Stop whinging. Should we listen to every complaint about market price or should we listen to none?

khal's suggestions sound interesting but also complicated. maybe someone finds an elegant solution.

on the other hand - maybe moa is right and I have to see it more as a free market than a fair place to register domains when needed.

I will just buy up plenty of domains and sell them for smaller price than the bad guys would (900% profit instead of 1000%)

[EdwardBlack]

Hi!

Edward Black here (Edward_Black on IRC, aka "the Tenebrix econ guy" aka "that stalker" )

First and foremost, some things I'd like to make explicit regarding my assumptions: I do not think crypto-coins are "currency", I happen to think they are speculative goods with notable features that make them a decent medium of exchange in some circumstances (examples of somewhat reminiscent entities among the usual physical commodities are diamonds and precious metals). Trading behavior of cryptocoins is definitely that of a speculative commodity of some sort.

Now, with that in mind, one could propose that primarily, the price of cryptocoins is affected by the following factors: mining user-friendliness and difficulty, coins retained long-term via certain kinds of services (good example - online wallets with instant internal transactions, which can tie up a lot of coins for a long time), coins lost irrevocably, and hoarding by users.

What merged mining did was introducing a lot of miners with no desire to hoard, thus creating a kind of "namecoin oversupply" which will likely remain in effect for quite some time.

The most obvious way to deal with that is to cut supply, which is exactly how BTC prevents issues that might (but are not guaranteed to) arise from constant steady creation of a speculative good.
However, I find that dangerous due to the fact that it introduces significant shocks to the market and could cause excessive miner exodus, as well as compromise seeming retention attractiveness should the market fail to adjust in a miner-friendly manner (it may very well be that profitable mining for fees is unsustainable), and induces fairly unwelcome changes to miner incentives even in a best case scenario (cartelization)

That's why I prefer an approach that counteracts unmitigated production of new cryptocoins by means of increasing irrevocable coin loss (which is the point of "fee furnace" I suggested for Tenebrix alt-chain), and namecoin is in a uniquely advantageous position since it has exactly such a mechanism and a strong incentive for people to throw coins at it, namely, the destruction of .bit registration fees.

My proposals are thus

1) remove subsidy halving (we shall prevent uncontrolled expansion of number of coins in circulation by burning them)

2) get rid of registration fee reduction

3) set registration fee to a fairly significant value

4) create a mechanism for automatic adjustment of registration fee (more on that below)

5) make renewal fee a "burned" fee and bring it significantly above zero

6) create a mechanism for automatic adjustment of renewal fee (more on that below)

7) in the (upcoming?) domain name transfer mechanism, introduce a transfer fee (also "burned") and make it to always be 50% of current registration fee.

This approach will aggressively solve issue of excessively cheap registration without causing any concerns as to possibility of registration becoming too expensive or too cheap, and also nicely gets rid of risks associated with miner subsidy cuts while at the same time not introducing risks of price collapse due to "eternal" constant production of the speculative good in question.

Now, all of this brings us to the following question: how to automatically adjust registration fees and renewal fees, given that having devs punch in new values is kind of clumsy, and blockchain can't go crawl exchanges (at least not without very radical makeover)?

Fortunately, namecoin is in a unique position of being able to tweak those fees based on data that can be found in the blockchain, as long as the following assumptions hold:

1) decrease in price of namecoin will proportionally increase the number of squatters and opportunists rushing in and buying up domains

2) increase in price of namecoin will proportionally increase the number of people who refrain from registering due to registration being too expensive for them.

3) decrease in price of namecoin will ensure that almost everyone who has a .bit domain will refresh it unless decided to abandon NMC completely

4) increase in price of namecoin will proportionally increase the number of people who have to refrain from refreshing and will thus loose the domain name due to refresh being too expensive for them.

Data regarding number of registrations and refreshes per a given period (measured in blocks) could be lifted from the blockchain, and existing data regarding new registration rates and normal renewal rates of various national TLDs is likely available and can be used as general "ballpark estimate" guideline.

I propose that you guys discuss this, and I will weigh in further slightly later.

[foreverD]

Here are some possible solutions to this problem:

1. Temporary update rights transfer to another address, while keeping ownership rights. This will encourage people to rent out the domains they are sitting on, rather than simply parking them.


2. BTC/NMC trading hard-wired into both block chains though special contractual transactions. For example:

"bid transaction": send 100 NMC to address A where they remain locked for 4 weeks UNTIL 10 BTC are sent to address B, after which the 100 NMC are sent to address C.

The bitcoin protocol allows for this in principle. See "Trading Across Chains" example:
https://en.bitcoin.it/wiki/Contracts


3. As above, domain name/NMC contractual transactions. This would allow people to offer their domain names for sale directly in the block chain with zero trust. The advantage is that you don't need to go searching for the owner if you are interested in a name that is already taken.

4. Higher update fees.

[phelix]

People working on an alternate p2p dns system (dianna) throwing fud at namecoin got me to look into the issue once again. I think I understand the original intentions better now and want to sum up.

To Khal's post above I would like to add another aim: keep the miners mining


Fees
There are two types of fees:
* network fees for name_new and name_firstupdate
(being destroyed which is currently not a problem at all)

Initially the network fee for name_firstupdate was high to give late adopters a chance. It continually decreased to the current stable amount (0.01NMC).


* transaction fees (going to the miners just like bitcoin)

For every transaction you can decide on the fee to pay. The miners decide if they include your transaction or if they consider it too low. At the moment 0 transaction fees still work well. This means the miners=pools don't care or still consider the block generated coins sufficient.

These fees are intended to keep the network from being spammed and pay the miners - just like bitcoin.

From what I understand the recent dust transaction attack was possible because the pools do not require a fee. That's why the mininput option was added that ignores very small transactions.

I expect much less domain transactions per user than with bitcoin so the network should have lots of reserves in terms of wallet and chain size.

NMC Value
Unlike most other cryptocurrencies namecoins have some inherent value as they allow domain registration and management.

This value is determined on the exchanges by supply and demand.

Putting it all together
Depending on the fiat/btc value of nmc the miners choose transaction fees to maximize their profit and limit chain size (this is in their own interest as it is on their servers - this is a critical point).

So we should get an equilibrium between mining effort payed with block generated namecoins + namecoin transaction fees vs. the value of domain handling.

[phelix]

previous post updated. had to get this out of my head

[khal]

Due to the recent fud about namecoin, i take some time again on this problem (not really urgent), after having done some work on dns suffixes.

EdwardBlack has interesting thoughts and a clear approach of the problem. Some comments :

That's why I prefer an approach that counteracts unmitigated production of new cryptocoins by means of increasing irrevocable coin loss (which is the point of "fee furnace" I suggested for Tenebrix alt-chain), and namecoin is in a uniquely advantageous position since it has exactly such a mechanism and a strong incentive for people to throw coins at it, namely, the destruction of .bit registration fees.

My proposals are thus
1) remove subsidy halving (we shall prevent uncontrolled expansion of number of coins in circulation by burning them)

Not a real obligation, but makes things more linear and simpler.

To Khal's post above I would like to add another aim: keep the miners mining

Previous suggestion match this idea.

2) get rid of registration fee reduction

Old formula need to be buried ! :p

3) set registration fee to a fairly significant value
4) create a mechanism for automatic adjustment of registration fee (more on that below)
5) make renewal fee a "burned" fee and bring it significantly above zero
6) create a mechanism for automatic adjustment of renewal fee (more on that below)
7) in the (upcoming?) domain name transfer mechanism, introduce a transfer fee (also "burned") and make it to always be 50% of current registration fee.

Renewal and transfer is the same command (and was available since the beginning). So, same price.

This approach will aggressively solve issue of excessively cheap registration without causing any concerns as to possibility of registration becoming too expensive or too cheap, and also nicely gets rid of risks associated with miner subsidy cuts while at the same time not introducing risks of price collapse due to "eternal" constant production of the speculative good in question.

Now, all of this brings us to the following question: how to automatically adjust registration fees and renewal fees, given that having devs punch in new values is kind of clumsy, and blockchain can't go crawl exchanges (at least not without very radical makeover)?

That's the good question

Registration fees :
I tested several formula (pages and pages full of numbers, awful...) to estimate what would be the ideal registration fees, from really complex (take into account number of not lost coins since the beginning, etc) to simple, and i think the simplest is the best.
Here is the proposal :
* registration fees is equal to number of coins generated per block divided by average number of domain operations (registrations + updates) on last 2096 blocks divided by a ratio of 2
* number of domain operations is an average on last 2096 blocks to avoid too big variations
* ratio is here to allow more operations in this period than in the previous one (even if there are more than 2 millions of nmc not used for domains).
* registration fees is calculated when difficulty change (namecoin will need to store it)
* registration fees can't increase/decrease by more than 30% each time, to avoid too big variations and refused registrations when price change (software may tolerate the old price during some blocks or the whole period)


Here is what price would be with this formula (with 1000 blocks[max 15% change = 1.15] and 2096 blocks [max 30% change = 1.3]) :
https://docs.google.com/spreadsheet/ccc ... zcmc#gid=1
- domain operations include name_firstupdate + name_update
- i didn't apply a 50% price for renewal or registration, but 100%
- you can change ratio and limit to plat with the price

Network fees :
- another problem to solve too ? Or should we rely on miners to updates min fees + share their value ? (i think this is not really a reliable system :p)

Any comments ?