NXT :: descendant of Bitcoin - Updated Information

[Anon136]

Edit: I've seen the date of ur post, it had been posted 1 day before Nxt released. Did u write about ur invention somewhere else or earlier?

No that was the first time and I wrote about it the same day that i thought of it. But you say that I posted it 1 day before nxt right? that technically makes it my invention right?

Right. And technically this makes me to suspect that u r BCNext.

Edit: I'd like to hear ur opinion regarding https://bitcointalk.org/index.php?topic=364218.0

haha no I'm a scrub programmer. my highest programming accomplishment to date is chess in the console with no AI. I'm much better at conceptualizing higher level abstractions than the logistics of implementation.

actually this is a really common thing. inventions are more often than not invented by a half a dozen people all over the world at the same time. we probably thought of it at the exact same time and i just wrote about it first. here is wikipedia on the phenomena http://en.wikipedia.org/wiki/Multiple_discovery

ill check it out.

[1714138465]

1714138465

[1714138465]

1714138465

[1714138465]

1714138465

[Come-from-Beyond]

Will you be posting instructions on how to assign alias before block 22000?   

thnx 

Yes.

[Anon136]

can you link me to the original source where you got this information. i really want to read more.

Other guy has answered on nextcoin.org about that.

I think I'm starting to get it. Is it true that the generationSignatures are totally deterministic? If you wanted too you could spend computing power to calculate who would win the right to author blocks far off into the future?

[Come-from-Beyond]

I think I'm starting to get it. Is it true that the generationSignatures are totally deterministic? If you wanted too you could spend computing power to calculate who would win the right to author blocks far off into the future?

Output of next signature depends only on input of previous one. For example:

F(x) = x*2

Then we have such a sequence of outcomes: 1, 2, 4, 8, 16, 32...

Transactions included into a block can't change it, there is a special "blockSignature" for them.

[Come-from-Beyond]

I'm preparing version 0.4.0, if noone reports a bug in alias system within 2 hours, I'll publish it.

[laowai80]

I'm preparing version 0.4.0, if noone reports a bug in alias system within 2 hours, I'll publish it.

And just to confirm, 0.4.0 is to be deployed in real network?

[Anon136]

I think I'm starting to get it. Is it true that the generationSignatures are totally deterministic? If you wanted too you could spend computing power to calculate who would win the right to author blocks far off into the future?

Output of next signature depends only on input of previous one. For example:

F(x) = x*2

Then we have such a sequence of outcomes: 1, 2, 4, 8, 16, 32...

Transactions included into a block can't change it, there is a special "blockSignature" for them.

Great! im definitely beginning to understand.

So next question. What mechanism prevents people from purposefully creating accounts with public keys which will allow them to author an upcoming block?

*edit* there is definitely a tip in this for you if you have good answers to all of my questions 

[Come-from-Beyond]

I'm preparing version 0.4.0, if noone reports a bug in alias system within 2 hours, I'll publish it.

And just to confirm, 0.4.0 is to be deployed in real network?

Yes. It will be invalidating all AS transactions till block 22000.

[laowai80]

Yes. It will be invalidating all AS transactions till block 22000.

Ok, got that.

Question on fees for aliases.
What should we set for a fee? Would it matter if someone was a few seconds later in generating an alias registration transaction but paying a higher fee, would that get them a priority? Is there a default fee?

[Come-from-Beyond]

So next question. What mechanism prevents people from purposefully creating accounts with public keys which will allow them to author an upcoming block?

Cunicula (as attacker) and BCNext (as defender) came to the following algo:

1. Brand new accounts must wait 1440 blocks before they r given permission to forge blocks
2. Old accounts forge blocks as if their balance doesn't include coins received in the previous block

Amount that is used for forging is called "effective balance".

[Come-from-Beyond]

Question on fees for aliases.
What should we set for a fee? Would it matter if someone was a few seconds later in generating an alias registration transaction but paying a higher fee, would that get them a priority? Is there a default fee?

NRS picks transactions with higher FEE/SIZE ratio in the 1st place.

[Anon136]

So next question. What mechanism prevents people from purposefully creating accounts with public keys which will allow them to author an upcoming block?

Cunicula (as attacker) and BCNext (as defender) came to the following algo:

1. Brand new accounts must wait 1440 blocks before they r given permission to forge blocks
2. Old accounts forge blocks as if their balance doesn't include coins received in the previous block

Amount that is used for forging is called "effective balance".

is it really that difficult to calculate out 1441 blocks ahead? it doesn't sound difficult. If it is so difficult now what happens when computers become faster and more powerful in the future? is the plan to just fork it from time to time?

it seems like if the generationSignature was additionally a product of the hash of the public key of the account that authored the previous block than that would add a measure of security for little additional cost. as in, unlike including phony transactions to get the result you want out of the digest of a whole block, it would be difficult to coax the result you wanted out of a public key funded with a large amount of cash 1441 blocks ahead of time.

[pinarello]

I'm preparing version 0.4.0, if noone reports a bug in alias system within 2 hours, I'll publish it.

Are you a human? Do you ever sleep?

Great cant wait !


Pin

[Come-from-Beyond]

is it really that difficult to calculate out 1441 blocks ahead? it doesn't sound difficult. If it is so difficult now what happens when computers become faster and more powerful in the future? is the plan to just fork it from time to time?

What should be calculated ahead to be able to generate all blocks?

[Come-from-Beyond]

I'm preparing version 0.4.0, if noone reports a bug in alias system within 2 hours, I'll publish it.

Are you a human? Do you ever sleep?

Great cant wait !


Pin

U still have to wait till block 22000 before u r able to register names.

[Anon136]

is it really that difficult to calculate out 1441 blocks ahead? it doesn't sound difficult. If it is so difficult now what happens when computers become faster and more powerful in the future? is the plan to just fork it from time to time?

What should be calculated ahead to be able to generate all blocks?

My understanding is that the next generationSignatureB is a digest of generationSignatureA and generationSignatureC is a digest of generationSignatureB ect... So what i mean is, what stops an attacker from calculating generationSignatures 1441 steps into the future and then creating and funding account that will win the right to author that 1441st block. It should in theory be rather simple to generate 1441 signatures if it is a purely deterministic process right?

[Come-from-Beyond]

My understanding is that the next generationSignatureB is a digest of generationSignatureA and generationSignatureC is a digest of generationSignatureB ect... So what i mean is, what stops an attacker from calculating generationSignatures 1441 steps into the future and then creating and funding account that will win the right to author that 1441st block. It should in theory be rather simple to generate 1441 signatures if it is a purely deterministic process right?

U must predict amounts on other accounts in 1440 steps, coz time when u generate a block depends on ur effective balance. It's possible only for empty blocks. If someone's account becomes larger then he will get chance to inject his block into ur chain completely ruining all ur computations.

[Anon136]

My understanding is that the next generationSignatureB is a digest of generationSignatureA and generationSignatureC is a digest of generationSignatureB ect... So what i mean is, what stops an attacker from calculating generationSignatures 1441 steps into the future and then creating and funding account that will win the right to author that 1441st block. It should in theory be rather simple to generate 1441 signatures if it is a purely deterministic process right?

U must predict amounts on other accounts in 1440 steps, coz time when u generate a block depends on ur effective balance. It's possible only for empty blocks. If someone's account becomes larger then he will get chance to inject his block into ur chain completely ruining all ur computations.

oooh so the generationSignature is a digest of the previous generationSignature + the time-stamp of the previous block?

[Come-from-Beyond]

oooh so the generationSignature is a digest of the previous generationSignature + the time-stamp of the previous block?

No. It's digest of prevGenSignature only.

How determined the generator (account) of the next block:

1. X = digest of prevGenSignature gives 256 bits
2. First 64 bits of X is a HIT
3. Target = BaseTarget (which is the same for all accounts) * Balance * Time_since_previous_block
4. If HIT < Target then next block can be generated

Edit: Target grows each second until it "crosses" HIT of one of the accounts

[xibeijan]

... planning to get rid of the interface completely.

- very sad.

The browser (https) access and the absence of stand-alone client is the most distinguishing (and attractive) feature of Nxt.

The browser access successfully works on smartphones right now.

Hope BCNext will change his mind regarding browser client.

I don't like the java interface at all and would prefer a stable, fast, clean native interface.

me too.
i want a client which is lean, small, fast and cross os, running
without installation from my usb-stick on windows, linux or mac,
able to connect to my private local node, my public net node and
even to other secured nodes later without any stored browser
history and security concerns like this, just by one click from
any place.

i want to manage my account, handle shares on the asset exchange
and use all the cool features nxt is providing soon without
leaving the client workspace, just like a central application hub.

i also want to see all quotes for other coins, best in near realtime
and trade them with advanced tools and technical analysis including
candlestick charts. auto-notifications on price-actions and bot-trading
with backtesting tools are also welcome.  

i want a client which never needs updates, because this is built in
and it would also nice to add special plugins with a single click and
use them without restarting my os or even the client itself.
all this should fit into 1mb to send the client via mail to friends
without bothering them by complex multilayer installations and make
them start nxt easy.
  
well, we don't have it right now but we could have it soon. ;-)

http://87.230.14.1/charts/01_screen.png

(early prototype running on my win7 vm)

Looks like a trading platform to me, and looks great.