NXT :: descendant of Bitcoin - Updated Information

[GCInc.]

209 blocks were generated by accounts with chance to generate < 1%

I'm within that list. How can we verify our account has created a block? There is no transaction related to that visible on the client. The block probably did not contain any transactions?

We would like it to be linear in account holdings, so there is no benefit from combining or dispersing your holdings.

Well done. I don't see logic in not fixing this.

[1714885226]

1714885226

[1714885226]

1714885226

[1714885226]

1714885226

[1714885226]

1714885226

[1714885226]

1714885226

[Come-from-Beyond]

Here is the mathematics behind the proof-of-stake generation rate.
Basically it says that the generation rate is proportional to the log of account coin holdings.
It is optimal to have 1 coin in each account when mining.
We would like it to be linear in account holdings, so there is no benefit from combining or dispersing your holdings.

http://www.scribd.com/doc/187987757/Nxt-Problem-Cunicula

I will postpone any further work in documenting Nxt until this is fixed.
Conjure up the developer and I will help him come up with a fix.

Edit: Fix is actually super simple. Just remove age from the equation and make the target depend only on the number of coins in the account.

I lost u at the very beginning. To continue further reading I have to know:

1. Do u decribe a Poisson process?
2. What does "age" mean?
3. Why do u use approximations?

[kwest]

Hmm... strange. Am I seeing other people's transactions in my unconfirmed?
I just saw a 30 million transaction from/to completely unknown addresses, and a 5000 transaction, in my unconfirmed list.
Those are not mine, and non of the addresses are mine. So why am I seeing them?

The transaction ID of the 5000 one is: 4186390315154153878

[Come-from-Beyond]

I understand that I can now claim my NTX, but I don't understand how to do that.
 

Could someone please explain or at least point me in the right direction (link)?

Thanks
 

Send 0.0001 BTC to 1BCN1ugdKdWd9pQ8Am9hMhtHZfmbXzxE8a with attached Nxt account id. Important: Send it from the same address that u used to send bitcoins from.

[Come-from-Beyond]

Hmm... strange. Am I seeing other people's transactions in my unconfirmed?
I just saw a 30 million transaction from/to completely unknown addresses, and a 5000 transaction, in my unconfirmed list.
Those are not mine, and non of the addresses are mine. So why am I seeing them?

The transaction ID of the 5000 one is: 4186390315154153878

It's a list of ALL unconfirmed transactions.

[kwest]

Hmm... strange. Am I seeing other people's transactions in my unconfirmed?
I just saw a 30 million transaction from/to completely unknown addresses, and a 5000 transaction, in my unconfirmed list.
Those are not mine, and non of the addresses are mine. So why am I seeing them?

The transaction ID of the 5000 one is: 4186390315154153878

It's a list of ALL unconfirmed transactions.

Ah.. OK. Did not know that.

[Come-from-Beyond]

2) Age is the number of confirmations on the coins used for mining. The target is proportional to coin-age. So it is difficulty/ ((number of coins)(confirmations on coins))

A fix for the issue is to change the target to difficulty/(number of coins).

Anyways, if you don't understand than just simulate it. It is easy to do in excel, matlab, or whatever.

Target is not proportional to coin-age.

BCNext wrote:

After thinking about the mining algorithm I came to conclusion that original proof-of-stake used by PPC and NVC is a bit flawed.  Bob could accumulate small amounts on different accounts during a long period of time and then attempt a 51% attack.  Artificial limits like max 90 days don't seem to work as intended.  Nxt will use a different proof-of-stake approach, I need time to nail some details and then I'll post them here.

After that he devised such an algo:

1. Target = BaseTarget * Balance * Time_since_last_block
2. Hit = First 64 bits of SHA256(Sign_Curve25519(PreviousBlockHash))
3. If Hit <= Target then a block can be generated

Thus

Time_when_next_block_can_be_generated = First 64 bits of SHA256(Sign_Curve25519(PreviousBlockHash)) / (BaseTarget * Balance)

PreviousBlockHash and BaseTarget are the same for all miners. Balance is fixed. There is no age here.

A fix for the issue is to change this to
target=difficulty/(number of coins).

We already have this!

[cunicula]

I think you are right. Good news.

False alarm. I was basing this on an older description of Nxt. The code already removes effects of age on hashing power. Going to delete older posts.

Specifically, I was confused by this:

Alice has 2500 nxts on her account. Last time she found a block 4 days ago. Her money is like a mining rig with hashpower equal to 2500 * 4 = 10000 GH/s.
Bob has 1000 ntx on his account. He was on vacation and hasn't opened his account for 20 days. His money is like a mining rig with 1000 * 20 = 20000 GH/s hashpower.

But later BCNext says this (see below), which doesn't mention coin-age at all. So like you say he must have changed the spec. I was confused why he criticized PPCoin and ended up adopting an almost identical system. It would make a lot more sense if he dropped coin age altogether.

It's a good chance to tell the details...
Each block has "generationSignature" parameter. An active account signs "generationSignature" of the previous block with its private key. This gives 64 bytes which are hashed with SHA256. The first 8 bytes of the hash gives a number (I call it a "hit"). The hit is compared to the current "target" (64bit number). If the hit is lower than the target then next block can be generated.
The target for each account is proportional to the balance. Someone holding 1000 coins gets a 50 times bigger target than someone with 20 coins. Thus the owner of 1000 coins will generate 50 times more blocks than the owner of 20 coins (in the long run).
The target is not constant, it grows each second passed since the timestamp of the previous block. If noone generated a block on the first second then the target becomes 2 times bigger and so on. The base target is the target on the 60 second mark. If there is only a few active accounts then after a long time someone will generate a block because the target will become very big. If you open the client and log with any funded account you can see a ticking timer in BLOCKS widget. It shows when the target will become greater than your hit.

[klee]

I am glad cunicula is looking carefully on this!!!

[Come-from-Beyond]

I am glad cunicula is looking carefully on this!!!

Me too. We need more guys to peer review security, crypto and other aspects of Nxt though.

[lyynx]

What has become of the blockchain explorer? I have not been able to access for atleast 6 hours. Is it being updated or has there been a url change?

[opticalcarrier]

What has become of the blockchain explorer? I have not been able to access for atleast 6 hours. Is it being updated or has there been a url change?

I imagine that IP was getting HAMMERED by all our requests.  Would it be possible to build the common gateway into the .zip that is distributed out, that way if we can be our own block explorer (assuming the client is up to date on all blocks)

[SlyWax]

I am glad cunicula is looking carefully on this!!!

Me too. We need more guys to peer review security, crypto and other aspects of Nxt though.

How can we review security if we don't have access to code ...

[nexern]

What has become of the blockchain explorer? I have not been able to access for atleast 6 hours. Is it being updated or has there been a url change?

it's online. i am testing currently a "reincarnation service", to restart the node when freeze.
the new getState api is fine for this. until the client ist more mature, the explorer can be
out of sync with the blockchain from time to time.

[Come-from-Beyond]

I am glad cunicula is looking carefully on this!!!

Me too. We need more guys to peer review security, crypto and other aspects of Nxt though.

How can we review security if we don't have access to code ...

BCNext offered to reveal any part of the code. Part of the source is not enough to clone Nxt.

[nexern]

What has become of the blockchain explorer? I have not been able to access for atleast 6 hours. Is it being updated or has there been a url change?

I imagine that IP was getting HAMMERED by all our requests.  Would it be possible to build the common gateway into the .zip that is distributed out, that way if we can be our own block explorer (assuming the client is up to date on all blocks)

yes, not many mining on the node but many account and transaction request. some seems made by fetching bots.
therefore i will provide a json api, where users can connect to and get all the data directly.

[SlyWax]

I am glad cunicula is looking carefully on this!!!

Me too. We need more guys to peer review security, crypto and other aspects of Nxt though.

How can we review security if we don't have access to code ...

BCNext offered to reveal any part of the code. Part of the source is not enough to clone Nxt.

Where are those part of code ?

[Come-from-Beyond]

Where are those part of code ?

https://bitcointalk.org/index.php?topic=352286.0

[Pouncer]

I bought some nxt with a friend, 150k, and been trying to send him his half. Managed to send about 10k. After that each transfer got rejected. What could be the problem

Incorrect time, bad connections, low memory. Relaunch client and try again AFTER that transaction expire.

Cfb, what do you mean by bad connections? If you mean number of peers, I'd like to know if my location amongst my peers matter. In my case, sending from local machine has been impossible since Nxt was launched.

Let me explain. Say Alice & Bob each has 10 peers connected.
Alice is in Paris, surrounded by 10 peers all over Europe
Bob is in the southern tip of Chile, with the 10 peers way north of him
Is there any difference in success rate of sending for Alice & Bob?

- strange. I've sent 59 transactions so far, and have no such problems at all.
You should send couns only when your client is mining ("You can generate the next block in dd days hh hours mm minutes ss seconds" above the "Recent blocks [n]" vidget).

Curious. How many peers do you usually have? I've been trying to send from my local machine since the day Nxt was launched - zilch. Blocks all up to date but nothing goes through. But when I send through my VPS, 100% success. I think if you're at the fringes of your group of peers, it's really hard to send. Imagine trying to send from Invercargill or the Diego Ramirez Islands. I'm not connecting from there, but close!

[Come-from-Beyond]

Cfb, what do you mean by bad connections?

Maybe it's an incorrect word. Let me explain.

Imagine that u connected to a peer with almost no free memory. If u send him a transaction he may see it and even attempt to send back (and to other peers), but due to "not enough memory" error he won't be able to do this. Java virtual machine reserves not so much memory to Nxt by default and most of nodes face the memory error after a few hours of working. If u connected mostly to such peers then u get what I called "bad connections".

BTW, it's better to launch Nxt by adding -Xms and -Xmx options:

Code:

java -Xms512m -Xmx1024m -cp Nxt.zip Nxt