[Draft] A Comprehensive Upper Tier Model

Nagalim 2016-04-05 23:42:07 UTC #41

There is a discontinuity of the nbt devaluation when the nsr reserve gets refilled during a peg break scenario that I’m still trying to solve.

mhps 2016-04-06 01:15:21 UTC #42

I think we have to be more specific on what kind of black-swan event that is. If it is bitcoin price has an exploding appreciation so that most people are selling nbt and nsr to get into bitcoin, then the event is highly likely to be a spike event that reverses with the same force, when massive buy-orders sweep up all nubits in minutes. It would not be a good idea to burn nbt when declaring pet break. It’s better to let the LPs and gateways keep them.

If the event is something like authority cracking down on cryptos or nubits, the the course of action you propose is more reasonable. We should also have a recovery plan and let nubit spring back if condition is ready. Nu can be very resilient even a peg break happens. The more resilient it is, the more people will hold onto nubits, knowing that eventually it can be sold for $1.

Park rate is annual rate. If we pay 8.5% per month to the LPs, why can’t we pay 100% pa as park rate?

Nagalim 2016-04-06 01:43:06 UTC #43

Park rates can be taken advantage of indefinately whereas liquidity pools have targets or competition. Though in general I agree that park rates should be way higher and liquidity costs way lower than we’ve done so far.

The idea behind the park rate threshold was to give shareholders a way to signal or avoid nbt devaluation using a blockchain vote. I’d also be down with linking it to fees. In theory, we could make the park rate threshold obscenely high (100%) and when we want devaluation we could use the 1 day slot so it would only be high during the devaluation. Still, it acts as a minimum for park rates during devaluation.

There’s still the issue of creating more debt when the network is suffering, but the refill mechanism creates a way to bailout the debt over time so it may be considered to balance out in the long term.

mhps 2016-04-06 04:19:54 UTC #44

Right. We should urge voters not to vote for park durations that are longer than how often voters would vote. For example if voters evaluate and could change park rate once a month, then all park durations longer than a month should be 0.

It’s just borrowing money from nubits holders to help defending the peg. It’s not unlike selling NSR, which is borrowing from NSR holders with a belief that the cost will be paid back some day.

Nagalim 2016-04-06 05:03:17 UTC #45

There’s still a distinct difference with liquidity pools. Let’s take fixed cost as an example. Say a pool has a fixed cost of 1,000 NBT/year. If we implement say 10% park rates and 10,000 NBT park, we’ve already hit the same cost. If we have 100% park rates on 1 week intervals and one actor is holding most NBT in circulation, they can instantly attain 10,000 NBT just for that one week. Even if market pressures immediately die off (because they were the ones generating them), they still get their 10kNBT. And they can just rinse and repeat next week.

There’s an attack that park rates without other peg mechanisms opens up: If someone is holding a vast amount of NBT and park rates start increasing, they can wait and even sell a few NBT onto the peg in order to convince shareholders to vote for exhorbitant park rates. Then, they can put up every single NBT at the exact same rate. And that’s the attack, that shareholders are assuming NBT ownership is decentralized and parking will happen in a semi-continuous manner. If a collusion waits, then parks a large number of NBT at the same time, shareholders have no time to feedback even with short park intervals.

Also, I’m pretty convinced offering half the rate at twice the duration is always logically consistent. If people want to hold their NBT in the network for longer time for free, let them.

Nagalim 2016-04-06 05:45:08 UTC #46

Made the devaluation occur on buy side only (so many gateways and operators can stay up). Also, the ending is more complicated:

@mhps I’m still all ears about the 20% park rate trigger. Did you have a better number, or a smarter condition?

mhps 2016-04-06 06:34:40 UTC #47

You assume everything stays the same. It’s possible that after 1 duration you need those parked nbt to be liquid. Longer park durations just make the Nu liquidity system less responsive to control

i feel that we don’t understand how park rates affect liquidity well enough to determin a threshold rate because we don’t have much data. Using parked amount as an indicator might be better e.g. “if 30% of all circulating nbts are parked”. Still I have very little confidence in any number.

We need this

masterOfDisaster 2016-04-06 07:00:55 UTC #48

Right! What good is 100% park rate if still only few NBT are parked?

My gut feeling says the total parked NBT is indeed the better indicator than a park rate.
Looking at the percentage of parked NBT provides information that already includes market feedback.

Cybnate 2016-04-06 07:43:31 UTC #49

And that’s why we need flexibility and not a fixed cap as we can’t foresee all the scenarios leading to a peg break. We may try a cookbook for a number of scenarios but a motion to cover all scenarios is silly and a potential thread to the network in times of crises as Shareholders may hesitate to act swiftly and break a motion because a motion from the past said so.

TLDR; Guidance yes, descriptive no.

Cybnate 2016-04-06 07:46:04 UTC #50

Agree, and depending on the circumstance the park rate will determine how many NBT will be parked. Being able to set a maximum NBT to part would be more logical than only being able to set a percentage. We need a Dutch auction model here and selling NBT positions to park to the ‘lowest’ bidder.

Nagalim 2016-04-06 11:51:30 UTC #51

I dropped the fixed cap a few edits ago. How do you feel about the 25 mil/3 months dilution rate? Of course we can dilute more if we want, it is just a promise to our customers as far as what they can expect at a minimum.

Guidelines are fine, but without some actual numbers and thresholds they end up being meaningless. Jordan Lee attempted to keep everything flexible with regards to T6, but eventually needed to put in place motions with specific numbers and thresholds. A more complete and thought out version of these threshold motions is inevitable.

Nagalim 2016-04-06 12:37:49 UTC #52

also added this:
#Transaction Fees
When the standard is negative, shareholders are encouraged to increase network fees. In general, shareholders would rather implement Restricted Network Access and ostracize users from the network than dilute shares or lose the peg.

Nagalim 2016-04-06 12:46:12 UTC #53

My problem with this is that it defeats the original purpose of that trigger, which was to give shareholders direct control over whether or not a devaluation occurs. One could theoretically trigger this condition without even being offered park rates and paying very little in txn fees. It would just be an opportunity cost, which is a low price to trigger a false devaluation event.

mhps 2016-04-07 01:09:24 UTC #54

parked amount isn’t be t he only condition, isn’t this has to be met

jooize 2016-04-09 16:34:37 UTC #55

Can and should undevaluation happen to later restore the value of NuBits?

Nagalim 2016-04-09 19:14:44 UTC #56

I’m not sure I understand the question in relation to the bit you quoted, but this is the model I’ve been playing with for how devaluation ends:

The thing is, devaluation is actually quite easy to end. Simply start buying back NBT at $1 instead of $0.9 or whatever. Eventually, T1 and 2 buy sides will pick up again. Starting devaluation is more difficult because you have to turn stuff off (buy sides on pools) or update to a complicated feed.

jooize 2016-04-09 19:18:59 UTC #57

Ahh, okay!