SGB Reward Rate - what it means

[BillyOckham]

49 minutes ago, PunishmentOfLuxury said:

This is from May '21 but I think the process is the same. The highest 25% and lowest 25% of FTSO price estimates are discarded and those in the middle win the round. 

Thanks PoL but that doesn’t really seem to be how it works.

First up FTSO AU said that they don’t provide the rewards whereas the vid says they do after they take out their cut.  

Second up the fact that there seems to be so much disparity in returns doesn’t sound like a broad 50% swathe of providers getting the win.  And the relative vote strength doesn’t seem to correlate tightly with winning so that doesn’t seem (to me) to provide the explanation either.

So yeah, that’s how I thought it would work,  but the reality seems different hence my questions.

 

[billion]

On 10/4/2021 at 7:03 PM, PunishmentOfLuxury said:

The Reward Rate that you see on Flaremetrics.io is the level of water in each container, not the rate at which the level is increasing (though the latter could be derived from the changes in level if one wished).

So just to confirm, the gains are linear all week long? I was wondering if I can extrapolate from the info I have on https://flaremetrics.io/check-reward to estimate my end of week rewards. 

[PunishmentOfLuxury]

1 hour ago, billion said:

So just to confirm, the gains are linear all week long? I was wondering if I can extrapolate from the info I have on https://flaremetrics.io/check-reward to estimate my end of week rewards. 

Not exactly linear, but an approximate projection could be made, yes.

Edited October 5, 2021 by PunishmentOfLuxury

[brianwalden]

Nice work @PunishmentOfLuxury you made it to the big time:

 

 

[brianwalden]

15 hours ago, Seoulite said:

Presumably if bigger providers and exchanges start to join then this will become less of a problem.

But is it such a big issue I wonder? If every provider is basically copying from the binance website, is it an issue? The binance price feed is good enough for billions of dollars of trade every day. Is it necessary to reinvent the wheel?

 

14 hours ago, BillyOckham said:

Yes that’s the question I’ve been wondering for a while now.  This whole convoluted structure doesn’t really make sense when a simple index of reliable existing price feed APIs can achieve the same thing.

 

I think the real reason is to create some tokenomics attractors to the network.

On the wide variance on reward rates….   that is worrying to me.  It really opens the possibility of drifting to non ‘true’ prices simply to converge on the leadership.

Eg low reward providers notice that the leaders are always .05% higher than them,  so they simply add .05% to their own estimates to land in the reward sweet spot.   This has the potential to drift further and further from the real price as positive feedback loops veer everyone off course.

I could be wrong, I would like to be wrong.  I wonder what other networks do and use.  I know next to nothing about non XRP chains.

 

Ok, so first, I think the FTSO system is probably going to work out fine. The prices will be good enough. I think once the network is connected to enough exchanges through deposits/withdrawals to/from other networks and through the F-asset system and Flare Finance's token wrapping, etc., arbitrage will keep the prices in a healthy range.

You guys have identified two important issues, decentralization and feedback loops. Flare could just use Binance's or Chainlink or whoever's price feeds and it would be a whole lot simpler. They'd be good enough and Binance's exchange business, for example, is way bigger than Flare/Songbird - they have no real incentive to manipulate the price. But what happens if Binance goes down, as it is wont to do. Suddenly the whole network has no price feeds until Binance is back online. Decentralization isn't only about the accuracy of the data, but the consistency - there shouldn't be a singular point of failure.

The other issue is feedback loops. Think of a normal archery tournament where the target is clear - the arrows in the bullseye are the winners. The price feed system is like an archery tournament where the target is invisible, even if you hit it, you still don't know that you have. Instead the winner is the arrow that ends up in the middle of everyone else's arrows. The problem with this is that the next round, everyone is going to aim at where the last winner was. Without something to bring the system back to the true target, the system can just push itself off in it's own direction.

A recent example of this is Iron finance's TITAN stablecoin. Everything was going great until it wasn't. It lost its peg a little bit, which caused more people to sell it, which only pushed its price down more - the feedback loop drove it away from its peg rather than back toward it. The problem is that it only takes one of these death spirals for everyone to lose faith in the system. Think of what happens if something causes the price feeds to dive lower and lower once there's real activity on the network. Tons of people are going to get liquidated, the agents who are critical to the F-asset system will be among the hardest hit, millions of dollars worth of value on the network will evaporate. As I said, I don't think this will happen, but we shouldn't ignore black swan events when we know they're possible.

I dunno, the big hangup for me is that it seems like this system was designed with the idea in mind that there is no one price. There's the price from hundreds of different markets around the world. It seems like the idea was to have providers submit prices from all these different markets so they could be combined together to form the network price. But data providers aren't rewarded for providing the price in their local markets, they're actually kind of penalized for it. I think it's probably going to work fine in the end, but I'm the kind of guy who'd rather have the perfect abstract concept that's totally impractical in real life than than practical system that works fine for our needs, but can't be abstracted into a pretty little framework. Yes, I realize I'm weird.

[BillyOckham]

I think it’s really about trying to create tokenomics rather than a stable and reliable decentralised price feed.

if you just wanted that,  you could use the validator consensus to vote on feed APIs.  Create a consensus managed feed list of APIs and then median the prices. That’s decentralised and can be automated into an oracle.  But it doesn’t create excitement or rewards.

[ftso_au]

15 hours ago, BillyOckham said:

First up FTSO AU said that they don’t provide the rewards whereas the vid says they do after they take out their cut.

The video was made long before there was clarification on exactly how it all worked. 

The community wanted more information on how things would work, we made the video to the best of our knowledge with the information available at the time. 

The rewards are paid out directly from Flare, they’re never in our possession. We also receive our cut directly from Flare.

[ftso_au]

16 hours ago, PunishmentOfLuxury said:

This is from May '21 but I think the process is the same. The highest 25% and lowest 25% of FTSO price estimates are discarded and those in the middle win the round. 

Agreed. I’m not aware of any changes to the original distribution awarding the median 50%. 

Edit: There hasn’t been a change to the original process.

Edited October 5, 2021 by FTSO_AU

[BillyOckham]

6 minutes ago, FTSO_AU said:

The video was made long before there was clarification on exactly how it all worked. 

The community wanted more information on how things would work, we made the video to the best of our knowledge with the information available at the time. 

The rewards are paid out directly from Flare, they’re never in our possession. We also receive our cut directly from Flare.

I probably wasn’t clear.  My statement wasn’t a criticism of you guys.  Only trying to make the point that things were now different from the video.  And as you state, for perfectly understandable reasons.

I applaud the way you guys have engaged and educated the community and am slightly annoyed at myself if my words have seemed to criticise that.  
 

However…. the reward disparity has apparently surprised you guys as well as us.  The wide disparity seems inconsistent with the previously stated methodology.  So either our understanding of the consequences of applying rewards to 50% of providers is wrong, or the implementation differs from the understanding.

It could  be a case of unintended consequences…. the implementation is exactly as described and the mid 50% of providers are rewarded at their voting power pro rata.  And that system has a possibility of an emergent property of a couple of providers racing away from the others.

To be honest I’m hoping that some definitive documentation will be created or come to light.  Discord chatting is the least desirable way to disseminate knowledge.

 

[Seoulite]

8 hours ago, brianwalden said:

But what happens if Binance goes down, as it is wont to do. Suddenly the whole network has no price feeds until Binance is back online. Decentralization isn't only about the accuracy of the data, but the consistency - there shouldn't be a singular point of failure.

Right this is a danger if 100% of signal providers are sourcing from Binance, but I think that is unlikely to be the case and gets less likely the longer the list grows (and as exchanges like Bitrue join). 

Throughout all this in the back of my mind there is the assumption that the Flare team has gamed this out as much as possible in advance. I don't think they are just making up the central aspect of their network on the fly, and any discussions we are having now were likely had right at the conception of the project. 

That's not to say I have blind faith that nothing can go wrong. As you said, black swan events are always possible. But whenever someone pipes up (usually no one here, I mean relatively new, uninformed, or cantankerous posters) with 'uhhh there is this obvious problem that will destroy the whole system, I can't believe no one has thought of this yet...' then I roll my eyes.

Edited October 5, 2021 by Seoulite

[BillyOckham]

30 minutes ago, Seoulite said:

with 'uhhh there is this obvious problem that will destroy the whole system, I can't believe no has thought of this yet...' then I roll my eyes.

Speaking to the wider context of anything technical or scientific…

Yes it amazes me how ignorant people can be,  that they give something a moments thought and think they have found something that professionals who are doing it as their day job,  after years of study and training,  have not conceived of.

Having said that, I am not as certain as you that Flare have gamed every scenario.  Surprises could await in this new network.  Time will tell.

 

[djdhrubs]

Sorry for the noobish question but how exactly do you claim rewards. One of the places I'm delegating to is FTSO AU and it has a rewards page but I can't see where I can actually claim them?

Edit: just found out. It says to come back in 2 days to claim the rewards. Need to read up about this stuff.

Edited October 6, 2021 by djdhrubs

[BillyOckham]

1 hour ago, djdhrubs said:

Sorry for the noobish question but how exactly do you claim rewards. One of the places I'm delegating to is FTSO AU and it has a rewards page but I can't see where I can actually claim them?

Edit: just found out. It says to come back in 2 days to claim the rewards. Need to read up about this stuff.

Rewards run over an Epoch (a week).  Saturday to Saturday.  Your selection of providers is locked in at some random point between Thursday and Saturday.  Remains in effect for that epoch and onwards unless changed.

Rewards are given as SGB.  If you want to compound them just wrap them and they will be allocated as you’ve previously delegated.

Always leave a small amount of SGB unwrapped to pay transaction fees.  Claims, wrapping, delegation are all transactions and need a sub cent fee paid.

[jbjnr]

On the subject of rewards ...

I sold 50% of my SGB, so I would expect to only get 50% rewards on this epoch - however, I switched from the not so high performing ftso_au and ftso_eu to two of the top 3 performers and am getting almost 2x returns - so overall, I'd expect my rewards to be about the same as last time. I see that they are a little bit lower (so far), and this can be explained by me farting around and sending tokens between wallets, losing some rewards - and also, not having precise records of how many tokens were in each wallet during the last epoch.

Other explanations would be -

• more people are delegating this time, so rewards are being spread more thinly (this is the reason AFAIK)
• the reward pool this epoch might be smaller (don't remember how big it was last week)
• (none of the providers appear to be capped in this epoch, so that shouldn't be a factor, but might come into play in the future).

Is anyone collecting data on rewards, inflation amounts, number of coins delegated, distributions by provider etc etc over time so that as the days and weeks turn into months - we can look back and see how things are changing/developing.

If not - can this be extracted from the blockchain/history reliably? Anyone know? (reluctant to ping ftos_au as they already do to much for us and are overworked, but noone else seems to care ...)

[BillyOckham]

1 hour ago, jbjnr said:

I sold 50% of my SGB, so I would expect to only get 50% rewards on this epoch

I don’t think this is correct.  After the lock in is done, (sometime after Thursday,) that is the amount and the delegated ratio that will be used for rewards even if you no longer have the wrapped SGB in the account.