Yes probably. Although I don't know how that works.
Let's say we detect some people are exploiting the PoI algorithm to increase their PoI index. We find a solution, change the code (probably NIS code?) and then release an update.
The exploiting people of course don't want to use the new release and nobody can force them, right? (Doesn't my local NIS decide my PoI index?)
Ok so its not intended that the fees one pays are higher than the gain through harvesting. Good, I understood that wrong. Thanks for getting that straight.
Im really curious about how things will develope. I think its very hard to decide about details of the PoI algorithm because nobody knows how many transaction will be done, how the NEM price will develope... and what people try to do to find a way to make money.
Does it really matter?
If PoI really works, you are not going to make real money from harvesting unless you are seriously envolved in something different from harvesting (so you don't care about harvesting).
Person a of your example will see his importance decrease day after day because other people are making transactions while he is not. In the log run he will be losing money (electric power or vps hosting are costing him).
Oh the index decreases over time if no transactions are made? I didn't know that.
If it matters? Well that is the question I'm thinking about. Maybe it does (because it could happen that people exploit the system), maybe not.
The costs for electric power are not really a problem though: Harvesting doesnt need much CPU power and if the computer runs anyway it doesn't make a big difference if you are harvesting or not. Right?
Oh the index decreases over time if no transactions are made? I didn't know that.
Not just like that.
We have seen it testing alpha: when you make a transaction your importance increase (so everyone else's decrease).
If it matters? Well that is the question I'm thinking about. Maybe it does (because it could happen that people exploit the system), maybe not.
The details are not public yet so we don't really know. I can think some tricks anyway:
if you know someone is going to move a lot of coins at a certain hour, make an outgoing transaction just before him so you increase your probability of forging the block with his transactions and get his fees
give away money to most important addresses
The problem is: these tricks cost you money and don't garantee you win. Are they worth? Probably not.
The costs for electric power are not really a problem though: Harvesting doesnt need much CPU power and if the computer runs anyway it doesn't make a big difference if you are harvesting or not. Right?
May be, but nothing is really free.
Consider this: NIS is not going to make your PC save energy because it has to write a block on disk every minute.
Yes probably. Although I don't know how that works.
Let's say we detect some people are exploiting the PoI algorithm to increase their PoI index. We find a solution, change the code (probably NIS code?) and then release an update.
The exploiting people of course don't want to use the new release and nobody can force them, right? (Doesn't my local NIS decide my PoI index?)
I wasn't talking about exploitation. I meant if POI doesn't behave like we wanted it to. I meant finetuning.
If such an exploit would be detected and a new version was released there'd probably be a hardfork and anyone not updating would be left on the fork.
Talking about writing onto my hard disk once a minute, are we going to have a configuration file where I can specify where to keep the database?
It will kill my SSD in no time with 1 minute updates. 
there is already a config file IF you're running standalone version
@mixmaster, here's a short summary:
[li]when you make a transaction, you need to add fee[/li]
[li]when block is harvested, ALL fees from the block goes to harvester[/li]
[li](as pat said) amount of NEM has influence on PoI[/li]
[li]making transactions influences PoI[/li]
[li]PoI of all accounts (mind accounts NOT harvesters) ALWAYS sum to 1.0. That implies, that if you're NOT making transactions, your PoI score will slowly decay, even if you have large amount of NEMs[/li]
[li]you CAN'T tell, how much you will gain from harvesting, that depends on number of transactions in a block, additionally TXes with higher fee, have better chance to be included in upcoming block.[/li]
Does that makes things clearer?
@mixmaster, here's a short summary:
[li]when you make a transaction, you need to add fee[/li]
[li]when block is harvested, ALL fees from the block goes to harvester[/li]
[li](as pat said) amount of NEM has influence on PoI[/li]
[li]making transactions influences PoI[/li]
[li]PoI of all accounts (mind accounts NOT harvesters) ALWAYS sum to 1.0. That implies, that if you're NOT making transactions, your PoI score will slowly decay, even if you have large amount of NEMs[/li]
[li]you CAN'T tell, how much you will gain from harvesting, that depends on number of transactions in a block, additionally TXes with higher fee, have better chance to be included in upcoming block.[/li]
Does that makes things clearer?
Thank you very much!
So, was the importance score calculation algorithm already published somewhere? I'm curious if I can break it 
So, was the importance score calculation algorithm already published somewhere? I'm curious if I can break it :)
No.
You can always use this:
https://bitbucket.org/mstrobel/procyon/wiki/Java%20Decompiler
I can never use this, unfortunately. I'm a simple mathematician, wrote my last program about 20 years ago...
So, was the importance score calculation algorithm already published somewhere? I'm curious if I can break it :)
Hi again mthcl :)
We'll try to push whole WP, but I wouldn't expect it earlier than in 2 months.
I'll send you PM when it'll be ready.
So, was the importance score calculation algorithm already published somewhere? I'm curious if I can break it :)
Hi again mthcl :)
We'll try to push whole WP, but I wouldn't expect it earlier than in 2 months.
I'll send you PM when it'll be ready.
Hi gimre!
Well, sure. But if you want to share it with me earlier, feel free to do so. I've heard somewhere that the algorithm is somehow related to Markov chains, and MC's is exactly my area of expertise...