The convergent curve was a mechanic implemented in Steem's EIP (economic incentive proposal) Hardfork 21. I was super upset about even the announcement that this was happening, and you can find me crying about it all over the blockchain during that time.
June 2, 2019
The EIP introduced 4 new mechanics (all garbage):
- 50/50 curation
- Decentralized Dev proposal fund
- Convergent Linear curve
- Free downvotes
But the sky didn't fall.
Indeed, no one even remembers what it was like before the EIP. Making so many changes all at once was super risky, but it really all worked out fine, contrary to all my bellyaching.
In fact, the entire bloated EIP system we implemented was probably worth it just for free downvotes. This was the most risky mechanic implemented, and it paid off big.
FREE DOWNVOTES BAD!?
I was sure that users would start selling downvotes because otherwise that's just a free resource going to waste. Turns out, no one wants to sell downvotes because if you randomly piss off the wrong person oops you just started a flag-war. Also, no one wants to buy downvotes, thus their is no supply, demand, or market for them at all.
Downvotes are much more carefully measured than upvotes are, and users are more than willing to let them go to waste. I have never cast a single downvote in 18 months (maybe one because @hextech witness team told me to).
In addition to that free downvotes killed the entire bid-bot system, which was the claim from the beginning, but I was highly skeptical. Was happily surprised when it actually worked.
I believe this is a sign that the Hive network should be instituting a 25% free downvote pool elsewhere like with witness voting and proposal voting. This mechanic will much better equalize the system as we've seen it work wonders in regard to the financial part of the equation.
Curation is a completely broken mechanic, so literally doubling down and doubling curation was obviously a bad idea. I've spoken to this multiple times already: it is totally backwards to enforce frontend behavior with backend mechanics.
As this pertains to Hive consensus, forcing users into 50/50 curation assumes that permissionless frontends are automatically going to display content with the highest upvotes. Spoiler alert: peakd.com can show you whatever content they want. leofinance.io can show you whatever content it wants. The idea that stakeholders are "curating" anything is a complete farce, so enforcing that on the backend is foolish and naïve.
Luckily, @blocktrades is coming to save the day, and is basically gutting curation by 99% by flattening the curve for an hour or more, allowing curation to be modified by a brain-hack kickback like LEO.
To psychologically incentivize Hive/LEO users to upvote other people and encourage decentralization while building out trust-networks, we employ "curation". So on LEO, curation isn't curation at all, it's simply a flat out 50% kickback to the upvoter with no need to gamify a reward curve and be the first one to upvote something. The 50/50 curation problem will be solved next hardfork, and that's great. I imagine the curve will be flattened indefinitely in the future as all testnets like LEO have shown that this is the best way to play the game.
Hurray for @blocktrades!
Speaking of @blocktrades and the EIP, have you guys seen the proposal for a tipping bot on Hive?! Nice, we are really putting that dev money to good use.
$2000 a day for 612 days
Only $1.2M for a tipping bot?! Sign me up! Didn't @crimsonclad put a ton of work into this exact thing and nobody used it? Nobody needs this, this is blatant leeching of the Proposal fund, and everyone knows it. This is why downvotes on proposals need to exist.
I hear @blocktrades is upvoting this, which begs the question: @blocktrades, how much money are they going to kick to you in backchannels if this gets funding? Can I get in on this? When do I get to wet my beak on this action? Should I buy a couple hundred thousand more Hive so I can get my cut?
10% of all Hive inflation is going to the dev fund. I've yet to see it generate enough value to justify itself, and it would be better if we just pumped that money back into the reward pool so that all stake holders can control it again. Paying $50k to a person who got us listed on exchanges? Only for that person to leave the network immediately afterwards? Cool.
Insult to injury, rather than destroying the Steemit Inc ninjamine the Hive elite pumped 80 MILLION HIVE into a fund that they might try to leech as well through the proposal system. We shouldn't have it both ways. Either the fund should be funded by the ninjamine, or Hive inflation. There is no reason for it to be both except to exploit the pool.
Despite all my Bitching, I'm still willing to give the dev fund a chance. 10% inflation might be a very small price to pay, and even if 90% of the projects being funded are shit, the remaining 10% could easily pay for everything and more. The dev fund is an extremely sharp double-edged sword that I want to play with, but I also don't want to chop off a finger on accident.
And so we come to the convergent curve!
Took a while, but we finally made it: the point of this entire post. The convergent curve cuts all rewards on upvotes by 50% until they reach a certain threshold. The more upvotes a post/comment gets, the less one in penalized in terms of rewards at the end of the 7 day payout period.
This has the advantage of making it harder to self-upvote with a bot army. Theoretically, someone like me with 1000 Hive accounts could spread out their stake all over the network and post thousands of comments everywhere, upvoting each one for a little bit. This would allow me to 100% upvote myself constantly with phantom circle-jerk accounts and it would be very hard to stop me from doing this.
The convergent curve eliminates the above attack vector because my small upvotes will be cut by half value. This means if I had simply engaged in curation like the rest of the good-actors on the network I'd make the same amount of money, while at the same time building out my network, making friends, increasing decentralization, etc etc etc. The network is trying to incentivize good behavior by the penalties of the convergent curve.
Who does the curve hurt?
Literally everyone who doesn't make enough money on their posts to break the threshold. Also, anyone who makes money on a comment is getting hit basically for the entire 50% penalty. So who suffers from the convergent curve?
The curve siphons rewards away from poor people on Hive, and funnels those pennies into the pockets of more popular bloggers like me. #winning. However, I also make a habbit of upvoting comments on my blog. I think it's good for engagement and networking.
Therefore I am also being punished by quite a bit. I have to upvote users for 7%-10% of my upvote in order for said upvote to not get dusted (2 cent threshold). To me, this is worth it. But if the curve did not exist this cost would be cut in half and I'd only have to upvote users for 4% to give them the same reward.
Tradeoffs: Is the cost worth the gain?
Short answer: no, of course not. Now that we have a free downvote pool, it's much easier to stop people from exploiting the network. We don't need a convergent curve that hurts everyone on top of that. It's not good for decentralization or engagement. Luckily we have projects like LEO that have already removed the curve that fleeces the poors.
How dare you self-upvote 100%!
The convergent curve also exists with the assumption that users should not upvote 100% and earn passive rewards and ROI. This is also obviously false. Users should be able to do whatever they want with their stake, and I'm tired of hearing otherwise.
Do you know why I upvote other people? Because it has more value than self-upvoting. This is all the incentive anyone needs. Building a social network of trust is far more valuable than scooping some paltry upvote rewards. If other people think otherwise, let them self-upvote... and if you don't like it then downvote their asses... that's how this network operates.
I've already made it quite clear multiple times that we should be implementing inflation ROI to the savings accounts and using those same accounts to double for CDP collateral buckets that can mint HBD. Once we implement these mechanics, anyone looking to 100% upvote themselves will simply do so via the passive rewards of the savings accounts, and then the convergent curve will truly become fully antiquated.
It's time to undo the damage that has been done. It's time to cut ties with all this code that Steemit Inc. wrote and forge our own path with DPOS governance voting. #hiveisthebest
Shill: Upvote @hextech witness
Posted Using LeoFinance Beta