You are viewing a single comment's thread from:
RE: Time To Wake Up and Fix Steem's Voting Problem
If I vote myself a $400 reward, there is no mechanism by which that will reduce my personal Steem holdings by $400 or more.
If the order book doesn't change then, no matter how many more Steem someone receive, if there's no one to buy them, then in btc/fiat/etc, you haven't enriched oneself. That's my point or at least a gross simplification.
Even if everybody does this [self-upvote] 100% of the time (which isn't the case), it is far from clear that it would destroy the value of Steem.
When everyone votes themselves exclusively, then votes become akin to interest on SP. As more investors join, we'll get ever closer to such a situation.
As this situation progress, SP use would be reduced to witness voting exclusively.
I have no idea what you are talking about in terms of the order book. At any given time it isn't hard to sell $400 worth of crypto at close to the market price (almost any crypto). If I earn $400 worth of rewards that I can sell it is very unclear what mechanism exists which would cause my stake to devalue by $400, and there probably isn't one. If it is just general inflation, then it doesn't much matter whether whether I earn the $400 or someone else does, but my personal wealth would prefer that I do, obviously.
No. In addition add: bandwidth, store of value, plus STEEM and SBD (both interchangable with SP to some degree) can be used a means of exchange. As bandwidth, SP may be valuable for other applications such as SMTs, Steem Monsters, censorship resistant accountable social platform, etc. Speculation about future value of these or other applications may impute a significant current value.
If someone increased their Steem stake by 1% every day but the price of Steem goes down by 2% every day then this person didn't make any money in fiat/btc/eth term. Their wealth has decreased.
How much money someone makes is dependant on the demand.
The demand can't be taken for granted.
True. I agree.
And if they don't increase their wealth by 1% per day and the value of Steem decreases by 2% per day, then their wealth has decreased even more. Furthermore no stakeholders expect this to happen because anyone who did would have sold and would no longer be a stakeholder.
Again, there is no mechanism by which one individual stakeholder's personal reward stream translates directly into inverse price changes. This is literally impossible because there are more than 100 stakeholders who can each increase their stake by at least 1%. It wouldn't be possible for the value to decrease by more than 100 x 1%.
The price of Steem is going to do what it will do. The individual actions of any stakeholder have, at best, a tiny influence. Therefore people can self-enrich. Therefore, the system must be designed to be robust to individual stakeholder self-enrichment. It either is or isn't robust depending on how critical you believe the reward pool function to be. This applies approximately equally to n or n^2 or any other curve.
No single one, but if all rewards are allocated through proof-of-no-brain, it devalues the system. I say devalue the system, it doesn't mean the price will go down.
What devalue the system is independent of who receive the rewards, it's dependent on the distribution mechanism and how it's valued by the investors.
The cryptocurrency space has seen its share of irrational investment, where totally worthless coin like Bitconnect had a much bigger marketcap than Steem. They ended up in flash crash.
Steem has value where Bitconnect only had a semblance of value. My point with Bitconnect is simply to illustrate that markets can be irrational but most people, most of the time try to be rational and that's what this world is based on, rationality.
If I should join the discussion here, Steem is like any other cryptocurrency project, if there is no demand (buyers) its dead, we can get all the formula right but our actions on the platform matters as sentiment would determine the volume coming to the platform...
If people haven't notice we have lost trading volumes from Upbit, certainly witnesses boast about dumping on the market and power down from steem accounts flooding the markets and upvoting those comments ridiculously, well we managed to successful flag the traders out of the platform, I was really disappointed to see how many witnesses supported the action of one whale who himself is dumping and earning much more.... You have to follow the reddit threads and twitter traders and see how traders view this platform, we must change sentiment, every rally is greeted with dumps, Steem cannot maintain value and some folks does not understand the system isn't robust enough to deal with the abuse...What sense if any it makes with flooding the market with Steem and SBD, traders take note of these things, people talk about it among their trading group... If we don't change sentiment and contain a few actors SMT wouldn't be enough
If the price doesn't go down, or doesn't go down enough as a direct result of the action taken, then self-voting is successful self-enrichment. The math on this is simple: If I vote myself $400, then I'm $400-X richer, where X is the amount that my stake depreciates as a result. If X < $400 (and it appears to be far, far, less than $400 in practice), then self-enrichment is the most logical, rational action to take. This is true for both n and n^2.
That is what you are missing. The incentives are misaligned which makes the system essentially broken. Both n and n^2 are broken. True they are broken in different ways.
BTW, n^2 doesn't not enforce proof-of-brain. That is the whole problem. It enforces proof-of-large-wallet. Something that enforced proof-of-brain would be great! I would be all in favor of it. (Actually I do think there is something that might: reducing the cost of downvotes.) I don't see n^2 doing that.
It isn't sufficient to argue that n is no good. You also have to make the case n^2 is indeed good, or at least that on net n^2 is better than n. The evidence for that after a year or so of experience with n^2 is far from compelling, and there is no logical case for it either: Proof of large wallet does not imply proof of brain.
Also, the white paper (and steemit blog post, etc.) argument about simple interest is wrong in the presence of downvotes, as I suggested earlier. You can always upvote yourself, but that's not enough to actually get paid! If someone else downvotes you (because your upvote does not meet their proof-of-brain standards), you are no longer guaranteed simple interest (and indeed your valuable vote power has been wasted, a potentially-worse outcome than voting for someone else and just getting the curation reward). You are forced to, at a minimum, create-and-upvote something that meets others' proof of brain standards and is not downvoted.
This theory is largely refuted after a year anyway. A lot of rewards are recovered by stakeholders via self-voting, but not all. There is still a very significant quantity of voting on merit, including when I have voted for your comments in this thread (and you for mine, and finally both us voting for others). In addition, posts with high merit such as Steem Monsters and other significant projects, get large amounts of merit votes.
So we can see that linear weighting does not reduce to simple interest in practice, and likely never will. So yet again, the theory is wrong because it does not describe how actual humans act.
What we see is might even be interpreted as evidence of stakeholders regulating the size of the reward pool in a somewhat healthy way (giving out more when there is more merit to justify the inflation, less when there is less) rather than inflating just for the purpose of inflating.
True. I haven't claimed the opposite. The wealth is dependent on the demand if the demand doesn't change then this factor is moot but the demand always fluctuates over time.
Short time this is true, but the longtime life of Steem is dependent on the market's demand over an extended period of time.
Under linear rewards, those who exclusively upvote themselves can hope to benefit from the Steem inflation minus the witness rewards. There's potential for these people to be flagged but then their potential reward would return to the reward poo,l where they will be available to anyone, including the other exclusive self-upvoters.
Currently, some exclusive self-upvoters are getting flagged every day. Those who flagged aren't getting richer and their flags are getting proportionally smaller compared to the rewards.
Under superlinear reward, @freedom could create 10 posts, occupy the whole trending page, get much more than the yearly inflation but that would decrease the demand for Steem to the point of not being the most monetary rewarding strategy.
Or @freedom could look to reward the people who will bring value to Steem and make Steem the center of attention of a growing amount of people, making Steem ever more valuable.
If any of the whales would decide to give themselves too much of the reward pool, bigger whales or combination of whales would have the power to raise their flags against those actions.
What is important is not merely whether demand fluctuates, but whether and how much it fluctuates (or can be reasonably expected to fluctuate) as a direct result of an individual's actions. Fluctuation, in and of itself, does not tell us anything about the best actions to take. I could do something absolutely horrific qualitatively in terms of Steem's value, and Steem could still go up in value! I'm only small stakeholder, even if, relatively speaking, one of the largest. When it comes down to it, my little actions just aren't that important (and the same can even be said for @freedom, etc.)!
Long term is just the sum of a series of short terms. If you want the long term incentives to work you have to fix the short term incentives. That is a mathematical fact.
Not necessarily. We've already discussed this.
Or even if so, @freedom could easily create 100 sock puppet accounts, post cookie cutter content that appeared to be meaningful and upvote those instead.
But even if that weren't the case, @freedom is only about a 3% stakeholder (as far as we know). Whatever @freedom does even with n^2, still has limited effect on the overall value proposition (see above). The perfectly rational thing for @freedom to do is to maximize @freedom's wealth with self-enrichment, which is pretty much what @freedom has always done (first with dumb but exploitative curation bots, now with bid bots). If you don't think so, then you are arguing that someone who seems to be a pretty smart dude and objectively one of if not the most successful Steemian who turned nothing into $20+ million by behaving very smartly at every stage is and has been doing something very, very wrong. I don't think so.
My very serious suggestion to you would be to spend some time with a pencil and paper and work out some very specific scenarios in terms of hypothetical (but plausible) stake distribution and also hypothetical (but plausible) numbers on how much these individual decisions effect the overall Steem value. If you do this you will find that the way the numbers work out, it is almost impossible for self-enrichment to ever not be the best individual strategy regardless of the reward curve (for each stakeholder who isn't already almost completely disenfranchised by a heavily top-heavy weighting system).
If this has been done early on, the white paper and overall design would not have made such obvious blunders. Either no one did this, or they didn't care because the goal was to launch a coin with a compelling-sounding story to cash out on a big pump, whether or not the mechanism for what it claimed to be trying to accomplish actually worked.