I will try to make the case for stability of bids rather than continuous increase in bids. But my heart nor my mind have truly taken any one side yet and I’m enjoying the debate from both sides, it’s fascinating.
“Then when enough vault owners install the modified vault then the price increases a lot.”
“good non-abusing people have applied the patch to gain better rewards”
This is where the argument falls short.
There won’t be enough vault owners install it because it won’t give better rewards, it would give lower rewards.
Nodes will see the rewards are lower using this software than if they use a strategy aiming at median bids (ie their ‘true expected’ reward). So you’d be asking someone to get less reward for maybe some potential future result. But only maybe. If there are efficient operators trying to lower the bid to exclude others then maybe the reward will never increase, so bidding high is an extremely poor strategy.
The question I would want answered is “Will people install something without verifying that it will actually earn them more”? I want to restate: bidding higher earns less not more, at least in the short term.
It feels like the incentives to bid lower is being totally ignored. There’s a lot of good reasons to bid lower (discussed further below in this post).
I think there’s also some good case being demonstrated by @neo to include a bias in the reward toward lower bids, eg if NB is 100 and there’s a bid of 99 and a bid of 101 the 99 bid gets slightly more reward than 101 despite being equal distance from NB. Does this sound like it could help?
“I would do it too and I do not wish to abuse the network.”
You would earn less, so presumably this bidding is seen as ‘altruistic’ rather than competitive to some degree, which seems a strange form of altruism. You would also be encouraging less efficient operators to join, so the network would be slower to function. It’s a bit like the 51% attack in bitcoin, even if you could do it you’d be putting your investment at an existential risk so you probably never would even if you could. I dunno, maybe I’m naive, but it feels like bidding high is the worst strategy, not the best.
I think high rewards is not bad per se. But reward amount should always be able to be adjusted (up or down) to allow improvements in operations to be factored in. The bidding is a mechanism to introduce balance rather than to introduce movement. If a high bid has no reason to be countered by an equally low bid then rewards should probably move that way. But ideally, high and low bids should usually balance each other out. Extremely high rewards is not sustainable, and would likely be balanced in some way by lower bids. I’m open to further thinking on this though, not set in my opinion at all.
“encourages an APP to investigate the section details and share them.”
I think this is good, right? The point of bids is to encapsulate market information. So sharing bid information (both about the bid itself and why that bid is made) allows for more efficient bid discovery. I don’t see any reason to keep bids secret. But I do expect bids to be placed in a timely and strategic way (in the same sort of way traders place orders in an exchange orderbook), ie usually as opaque as possible to keep the privilege of information to themself as long as possible.
“If 100% of users decide to trick the network, they will trick it and milk it out of coins.”
What is the trick though? If everyone bids high then the trick would actually be to bid low so high bids are far from NB and rewarded less. So it’s tricks upon tricks upon tricks until eventually the optimum trick is to bid the true median. Does that sound plausible?
“If you do not think this [exploitation by bidding high] would happen”
But isn’t there an equally valid opportunity to ‘exploit the exploiters’ on the opposite side by bidding low? The point is maximum reward happens not at the highest bid but at the ‘most balanced’ bid, so even if you can trick your opposition on the high side then you get tricked on the low side then it’s tricks to and fro simultaneously and everyone is second guessing until you realise let’s just try to bid the median. Not to say tricks won’t happen, but the stable equilibrium state is designed to achieve balance, not pushing to the extremes. Can you imagine stability or is it me being idealistic?
“Parsec does not prevent people from coordinating their price bidding actions in any way.”
Coordination is fine. You can coordinate to raise the bid. But you can also coordinate to exclude competitors by lowering the bid. And you can also coordinate to screw the coordinators (by convincing them to bid in a way that gives them very small rewards but you yourself bid for maximum@median rewards). Coordination does not ensure uniform behaviour, but that’s what’s being claimed (everyone will bid high ie uniformity in bidding). So I guess there’s a possibility of confusion of terms here - ‘maximum bid’ as in ‘at the upper boundary of legal bidding’, and ‘maximum bid’ as in ‘the bid that should give the maximum reward which is at the median of the NB’. This could cause confusion! I don’t think it has caused confusion yet but wanted to flag it anyhow…
“How you would stop inflation to not reach 100% of available coins ?”
Good question and not as easy as it sounds. I think firstly the distribution should be scaled according to remaining coins, so 90% remaining means giving a lot of reward, and 10% remaining means give not much reward. But that’s pretty simplistic so I’m open to further exploration. Another thing to consider is to set an absolute upper limit on bids. Obviously a starting point would be no bid more than 2^32. Then iterating further no more than 2^32 / number_of_sections. Then further again no more than 2^32 / number_of_sections / number_of_nodes. And I guess we can iterate further but I’m not so sure how to establish the absolute maximum allowable bid…
“give me some bones to start putting flesh on”
- using measuremnts such as space / latency / bandwidth / node count etc to determine the reward leads to an unworkable incentive algorithm (a bold claim for sure).
- is there a way to discover the correct reward and storecost from ‘the users’ rather than from ‘the users behaviour as proxied via supply of storage’. Why use storage as the ‘voice’ for rewards and costs?
- how do you prevent users from saying ‘obviously you should pay us more’?
- how do you prevent users from saying ‘pay us very little so nobody else can compete’?
- the main achievement is allowing storage and bandwidth and latency and reliability etc to be managed by the users rather than the network. The network should be a lot more resiliant and able to reward a wide variety of node behaviour rather than only target behaviours around storage (ie we should try to avoid wrapping all behaviour under the blanket of changes-to-storage).
“PA gets us into game theory and nash equilibriums that I think better represent evolving towards more and more adaptive behaviors.”
Agreed. I think the benefit is the clarity of the adaptive behaviours, rather than the number of them. Both reward systems allow vast amount of adaptive behaviour, but the storage-as-a-proxy makes it harder to discover and know you’ve arrived at the beneficial or detrimental region of gameplay. Just my gut feeling on the game theory.
“If the door is left open for collusion, it will happen.”
Can people collude on rfc-0057-style storage-parameters-reward-algorithm? Is that collusion more destructive or less destructive than in the bidding collusion? I don’t have an answer but this is important to ask. It seems hdd manufacturers have a big head start for power in the storage-parameters-algorithm and are ripe for collusion…?! Just trying to work out which algorithm has the least-bad collusion scenario.
“When people realise that if enough vaults join a coordinated group then they can raise their bid prices then there will be one or more APPs written to facilitate this coordination.”
Is this similar to bitcoin where miners join the most popular pool because the reward is the most regular? Because that did happen with deepbit and ghash, and sure enough miners moved away from those pools. If collusion presents an existential risk to the network farmers will act to avoid it. I reckon there is strong historical precedent for collusion being avoided in cryptocurrency, but likewise strong precedent for collusion happening elsewhere and being a problem, so we get into ‘he said she said’ territory I guess.
There are two opposing forces of greed that ideally should balance each other. One is the ‘obvious’ greed which is to keep pumping up the bid to get more reward. The other one is ‘subtle’ greed which is greedy by exclusion, and requires lowering the bid. Subtle greed aims to a) exclude competition by removing inefficient operators and b) exclude the rapid creation of new currency by keeping the reward rate low. So the case for the ‘subtle’ greed is really twofold (but doesn’t mean it will necessarily be the stronger force). I personally think of these two greeds as ‘dumb’ and ‘smart’ greed, but maybe that adds unhelpful emotional bias.
Here’s some separate late-stage hypotheticals to mull over, would be curious to know which you think is most likely:
-
Polarisation happens ah-la trump/brexit and opinions are extreme but also very evenly represented. Almost half the nodes bid the minimum possible and almost half bid the maximum possible. The swing bidders hold all the power and the ‘winning’ opinion gets all the reward, the ‘losing’ opinion gets none. But the winners and losers flip very easily and very often, so the reward ends up pretty even over time. It becomes more like a lottery.
-
The obvious greed continue to push the price up until it’s permanently stuck at the maximum (determined by total sections and percent of unfarmed currency). Bidding below the maximum can only cause loss of reward, and there’s no way to bid above the maximu, so there’s no benefit to bidding below the maximum and it stays stuck there forever.
-
The sensible greed causes a few large operators to dominate farming. When they realise that very few competitors are farming and there’s no reason for further exclusivity, they raise the reward again to get more benefit from their equipment. Other operators start to become viable but they can’t start up instantly. By the time competitors actually start being rewarded the big guys start pushing the reward down again and the competitors never really got started before they’re unviable again. The cycle repeats constantly between a few big farmers.
-
On a single farmer level, consider a farmer that has operated for a long time and has farmed quite a lot of currency, and has decided they’re not able (or willing) to compete with the new farmers. Their equipment is getting old and is nearly unviable and going through another equipment changeover is too much to bear. Should they try to raise the bid so they can maximise the last bit of reward out of their gear, or should they lower the bid to try and increase scarcity? I think it makes most sense to lower their bid aggressively because any new currency they farm will be a relative small portion of their total rewards, so it makes sense to maximise the value of their existing rewards, which means increasing scarcity and thus a low bid.
