πŸ“–

Beanstalk University Class #58

Date
January 25, 2023
Timestamps
β€’ 0:00 Intro β€’ 2:06 What is the purpose of the Field? β€’ 7:17 How did ESD and DSD borrow credit? β€’ 24:24 Upgrades to the Field β€’ 31:25 What other improvements would Publius like to see? β€’ 37:01 Closing statements
Type
Beanstalk University

Recordings

Meeting Notes

What is the purpose of the Field?

  • Beanstalk is a credit-based system, the value of Beans comes from it’s creditworthiness. The purpose of the Field is to attract lenders to Beanstalk during the Seasons Beanstalk needs lenders at a reasonable price. The general purpose is to borrow Beans from the market. Publius adds that there are a lot of open questions around the field in practice

How did ESD and DSD borrow credit?

  • Publius does not fully remember how they worked, but they have a rough understanding of how they work. Basis had an auction, there would be a Dutch auction at a certain time. The protocol was willing to sell debt at any price point. For ESD and DSD the debt issuance was a function of the debt level of the system, the interest rate was on a curve. Both of these had their own inefficiencies that cause long de-pegs. Beanstalk creates Soil when Beanstalk wants to buy debt. Soil gives you Pods based on the current Temperature. Beanstalk pays back Pods on a first-in-first-out (FIFO) basis. The FIFO system incentives users to Sow earlier as opposed to later. Beanstalk issues Soil based on the DeltaB. The Temperature is based on the Price, the debt level, and demand for Soil of Beanstalk. Beanstalk does not look at time below the peg or how far off the price is from the peg, but this could be incorporated into Beanstalk.

Upgrades to the Field

  • There have been a lot of BIPs to improve Soil issuance and how the Temperature works. The most notable upgrade to the Field is the change that got rid of cumulative Soil. Cumulative Soil caused Beanstalk to issue Soil where it did not need to issue Soil. Currently, Beanstalk optimizes around debt levels during times of high debt. There is an improvement to implement a Dutch auction so that the first 25 blocks within a season will ramp up to the current temperature.

What other improvements would Publius like to see?

  • Publius thinks that magnitude of the de-peg should be looked into more for Soil issuance and Temperature. They think this is a marginal improvement and they do not plan on spending too much time on this. They go on to say they think there should be downward reflectivity when Beanstalk re-pegs. There are so some open questions about how Beanstalk measures demand for Soil.

Transcript

Okay. I think we can get started. Alex just asked him in private. Why not use the everyone tag to notify everyone? You know, the reason we decided not to do that as we think the. Everyone was just, you know, constantly calling everyone causes some fatigue or on the importance of notifications. Hence when I'm working on it or use it for classes and all meetings, hopefully there has been some time changes with class, but we will go back and it will be in schedule so that, you know, should make it easier and unclear for others.

You know, when when, you know, class and dormitory sickness also took us all with us on maybe listen to this recording. They said that you can always find events that are scheduled on the on the channel. That's sort of the the menu at the very top of the the SQL Server. All right. Well, one more thing. There are two channel discussions for two bits that are going to be proposed.

Those may be submitted with them. So that's the the basic forms budget and the set up step. One notable change for the first budget is that this time, as you know, for the for the first half. So it's not a quarterly budget. And I would encourage everyone again, who is with us today will be listening to this recording to join the conversation and, you know, have any any questions and ask them there.

There will also be office hours but are listed in the announcement channel. You can know what day is that'll be there. The team from the BBC will be there and happy to take any questions or answer any questions regarding to the to the budget or anything else. All right. How are you for this? Doing quite well. How are you?

Things that things are well here. Yes. I was thinking that maybe we can spend some time or dedicate this class to talk about the field in depth. And, you know, this is something that, you know, we've occasionally discussed, you know, throughout times on different classes, but maybe we can start from the very beginning and go on, you know, but starting off, maybe, what is the purpose of the field?

Why does the field exist? And being stuck? Being stuck is a threaded based system the value of beans? At the end of the day, from a theoretical perspective, our ultimate derive from the credit worthiness of being stopped. In practice, what that means is the ability for beanstalk to dampen the volatility in the price of beans downward is a function of its ability to borrow beans from the market, and Beanstalk borrows beans from the market in the field.

So the field is the Beanstalk credit facility and the purpose of the field is to attract lenders to beanstalk in the quantity and during the seasons that bean stock needs to at a reasonable price. Now there's a general open question as to the best way for beanstalk to price the debt or or price. It's it's it's pod issuance in practice, which is the temperature the interest rate for lending or sowing beans to be in stock is the temperature.

There is a there's probably a lot more optimization and theoretical progress that can be made in terms of how to properly set the temperature in general. But the the general purpose of the field is to for being struck to borrow beans not as efficiently as possible. This is the comment on the temperature optimization being ambiguous at the moment.

It's not to borrow as efficiently as possible because efficient is it's hard to it's hard to define at this point. Now, does it mean that Beanstalk should optimize for borrowing as quickly as possible at higher interest rates or in general, like, for example, right now there doesn't seem to be much demand to lend to bean stock at current at the current temperature now being Beanstalk could in theory increase the temperature very quickly until it finds borrowers.

But if there's no demand to lend to being stuck at any price right now, for whatever reason, then whenever the temperature is found where someone is willing to lend the bean stock, it may be so high that beanstalk stock issues too much debt where now it's unable to find the next marginal lender. And so instead of just optimizing for a quickly finding a borrower at any price, Beanstalk currently raises the temperature relatively slowly and is less concerned with immediately finding a borrower in practice than if it were raising the temperature much faster.

But then there's an open question as to what is the reasonable amount of time in terms of seasons or time in general and price action or deviation from the peg that is acceptable to the system. And how should the the incentive to lend to being stock change in time relative to it? So there's a lot of open questions in the field in practice.

But the you asked a general question, so you got a pretty wide ranging answer. But the just to summarize, the purpose of the field is for being start to borrow beans from the market. Okay. And yes, we we spoke out, spoke a little bit about, you know, the temperature and maybe the well, the details of or the or how the field functions.

But maybe we take it a step back a little bit before diving into these details. So as you said, Beanstalk is a credit system. Credit means your ability to borrow. And given that beanstalk as a STABLECOIN so it uses its ability to borrow to maintain impact. And this this idea or in the thought was experimented or done previously in in previous iterations.

And we want to talk about how the field became the field today. And maybe before that we can briefly discuss how or what different iterations were done previously with other protocols maybe on how they wanted to borrow to do the same thing, which is to maintain peg. And we can start maybe with ones that we're running. And, you know, we have is the industry which both of them, they also relied on their ability to borrow to maintain peg.

And you know, we saw we saw various results on it can produce. Maybe you tell us how it is. The idea is to try to borrow credit and then from those outcomes, what were the lessons learned? And then how is the field and a different. So I'm not going to make it up mad and I don't remember off the top of my head exactly how they worked.

But perhaps we should talk about theoretically the different models that have been proposed. Now, the the, the, the basic idea for a debt based currency really started with senior and shares. And there were a couple implementations and attempts at implementations like basis basis cash, ESD, DSP and the there were a couple different models that were attempted. I'll try to recall as best as possible, didn't didn't know we were going to get into the field.

That also didn't didn't do any refreshers on this before. Apologies, but the if I recall correctly, basis which was based on the senior IT shares model had an auction and they never launched. So it never it never was tested in the wild. But the concept ought to be that there were basically there was a Dutch auction on a regular interval and the system was willing to sell debt at any price.

And that model, in theory, would be going back to the open question on temperature optimization, would would optimize for selling enough debt all the time to keep a perfect tag or be able to return the price to its peg. And whatever interval that the the bond issuance was taking place at any price. And at least from our perspective, that would almost certainly lead to incredibly high debt issuance at certain periods of time that would likely cause runaway debt issuance and the system ultimately to crash and in addition to that, there's the in addition to the issuance of debt, there's a question as to the redemption of debt and generally prior to being stock, the debt

issuance was fungible. And so redemption around the redemption around repayment of the debt was typically subject to some race condition. Now, with ESG and DSD, the the the debt issuance was a function of the debt level of the system. And so there was some hard coded interest rate curve that dictated what the interest rate should be somewhat arbitrarily, which led to large inefficiencies in the debt.

And so that that was subject to a different problem. On the debt issuance side than the basis model, because whereas basis would overpay for debt issuance, ESPN, DSD would issue debt on an arbitrary curve such that they would there would be long periods of time where there was no demand to lend to the to the protocol because the interest rates at which it was willing to offer, it was just unattractive and there was no protocol made a way to adjust to market conditions at all in practice.

So that was on the debt issuance side. And then on the redemption side, ESD and DSD, if we recall correctly, were subject to raise conditions on chain because the debt issuance was fungible, which led to whales being able to redeem large amounts of debt and smaller, smaller players being unable to redeem at all. So it created a really horrible, horrible user experience and it like it was an unfair market in practice and that created large inefficiencies because smaller players didn't want to participate.

And ultimately these types of systems are dependent on a large set of users, not just whales. So those were the problems on the debt issuance side and redemption side for the two major models that were previously attempted prior to being stuck. And just to now talk about how Bienstock fits into that context, that they also have an expiry expiry date.

So when you had a debt, was there a time where they can redeem it and then it expires? I think in all cases prior to being stuck, there was some expiry on the debt, although I don't recall exactly the time periods for, for any of those protocols. But that's a good point. But they all had expiry on the on the debt issuance, Correct.

Okay. So fungible debt was fungible in the sense that it was on the same. As soon as you buy it, of course, you could have bought it at different amounts, but there was no sort of which were going to come into that. When did you buy it? Exactly. And then the other point was also the expiry the expiry of it.

And then yes, the expiry would make it such that it wasn't totally fungible, but just in terms of redemption, it was typically fungible. Just to clarify. Okay. All right. And now we want to talk about, you know, how there'd been some try to try to address this. And I guess the two the two main things that we want to talk about is, is the fee for so similar to the first one and first out.

And then the second part is, you know, controlling, as you mentioned previously, the way that the you know, the interest rate was was measured was dependent on the the level of the debt in the system. But now being sort of differently, can you maybe tell us a little bit, you know, how how does the field address those two things?

Well, the what in Bienstock, whenever Bienstock is trying to borrow beans from the market, it creates soil and for each soil, anyone that has a bean can lend the beans to into the soil or so the beans into the soil. One bean to one soil, and they get pods based on the temperature at the at the time that they sell on a that that harvester become redeemable for a bean on a first in first out basis and in practice because the distribution of new bean mints or being senior edge is determined by the the code.

So the distribution of the the bean seigniorage is known in advance given that the pods harvest on a first in, first out basis and at the time that you learn to be in stock, you know the place in line and you know the number of pods that you'll receive, you know, at exactly the bean supply at which your pods will become redeemable into beans.

And because beans are only minted when the price of a bean is above a dollar, the the the there is an expectation that when the pods harvest, they will be able to be sold for at least $1 of value. And so given the price of a bean and the temperature you know at exactly what bean supply, you can expect a return in price.

That return. Exactly. And the the interesting thing here is that ultimately the previous bond models that the other protocols attempted were copies of current debt implementations in traditional finance, because you would never go to a bank or a lender or excuse me, to a borrower and lend them money based on, you know, some sort of parameter of their business.

Like, oh, trust me, we'll tell you when our business is at a certain point where we'll pay you back. That's ridiculous. But because bean stock is a smart contract and it can codify its repayment of the debt based on some parameter, there's no need to make repayment a function of time and instead of time or an expiry existing on the debt and the debt being some sort of fungible issuance of debt or semi vulnerable issuance of debt bienstock issues, pods that are redeemable based on the bean supply, which is something that is enabled by smart contracts and it's the first in first out harvest schedule combined with the fact that the in practice the temperature

doesn't change particularly quickly, makes it such that at the margin, if you're thinking about lending to beans, stock, the the cost of someone else lending to you to be in stock before you relative to the marginal benefit of the temperature increase and over the next hour or day, a couple percentage points, whereas the temperature right now is over 6,000%, it is its marginal benefit for waiting, whereas the risk of someone else lending before you and you your expected return, a return parameter increasing and therefore your expected value decrease and creates an incentive at the margin to lend to bean start at whatever price you think is an efficient price, or at least in a relatively

efficient way. Again, arguably it could be further improved, but that's that's the general juxtaposition of how bean stock issues, debt. Okay. And guys share a bit from the basis white paper that there's just a basis was the protocol that was never launched by was the initial let's say white paper or the first credit based stablecoin that was written.

And then the expression was five years is the the expression of the bonds where around 90 epochs, which was around 30, 30 days. So, so probably first, then first out who buys first or who gets holds the bonds first. As you know, the first one gets paid. That gets paid out. And then the the second important point here is that there is not expiration or, you know, the the bonds in that case or the cod.

That's your hold Don't don't expired. Then when it comes to the how the interest rate is calculated, maybe what's of interest as is the idea of a death of a spider. And by that spider, I mean maybe the you know, the interest are going out of control. So when and previously as it was a measurement of the debts or the higher the debt, the higher the interest rate.

There is no such thing that's sucked. So so doesn't look at a certain thing. And the more that thing is, the higher the interest rate is. So, for example, maybe, you know, how far are you from Peg then then Bienstock increases the interest rate by a higher rate. The way that bienstock increases the temperature or which is the interest rate is there's on demand for soit, but that increase is limited.

It's a set either one or 3%, irrespective of how big or how long or how far the price of being is. Or how long has it been that there has not been demand of it. Is that correct? Probably as yes. So currently bienstock issues soil based on the the delta B, so the change in beans from PEG generally and it's slightly different when it's above and below peg, but if Bienstock needs a 100,000 beans to be bought on the market for the price to return to peg, there will be 100,000 soil is generally how it works and the the temperature on that on that soil is changing exclusively at the moment based on the price

or really the delta B over the previous season what the time and liquidity weighted average delta B over the previous season, whether it was above or below peg and the debt level of bienstock. And the concept is that as the as the debt level changes generally when it's particularly high or particularly low, the the debt, the change in temperature excuse me should should fluctuate in addition to changing based on the price.

And there's right now the system is divided into four debt levels or pod rate levels and two price levels above and below peg and then extremely low, reasonably low or relatively low, relatively high rate, extremely high. And the the change in temperature is a function of those two parameters exclusively. Now, as you were talking about mud, there is no notion of time below peg or above peg and time at a certain debt level.

And there's also no notion of magnitude in terms of price. And all of those could be potentially incorporated into the change in temperature. And the change in temperature could be a change from some arbitrary value to some relative value of the the overall temperature. But that's more likely to have some sort of exponential change, which is probably acceptable on the downside, but not acceptable on the upside.

Because, again, the goal is never to have a runaway temperature that creates a death spiral. So yeah, there's a lot of open questions as to what is the optimal set of things to consider in the in the function that changes the temperature and there's also a question as to even if we don't change the inputs, just the right way to to consider the current inputs.

But currently being stuck only only changes the temperature linearly. As you said, one or 3% based on the the delta B and the the pod rate, the debt level. Okay. So this was taken into account. It was put in in design. Once again, the idea of, you know, what is the kind of the that that should which is not C4 and one that doesn't expire and then the you know how the temperature is changed And back then it was the weathered launches and then, you know, there are sort of improvements that happen.

And as you said so initially the so-called issue soil, the soil can be bought for beans, you know, soya beans into the soil. And then you receive pods. And initially when beans are launched, there was no cap on the amount of soil that's available or assured by by being stuck. That changed then. That probably is. Can you maybe tell us why did that happen?

What was the experience that, you know, that that happened or, you know, we the protocol experienced and then an improvement in it was wasn't reduced. So the there were there were a couple of different bits that changed the soil and temperature issuance. But the most notable from a theoretical perspective was that previously there was cumulative soil, meaning that being stock, if it was below, if the Delta B was negative for consecutive number of seasons, that the soil would accumulate as a sum across multiple seasons up to some maximum percentage relative to the beans supply, and that resulted in a massive amount of soil issuance that was unnecessary to repay the stock.

And so being stock would then be above peg and have lots of soil, which would then lead to a significant amount of debt issuance that was unnecessary. And so there were bits at least one day put in place to change the the soil issuance where now it's not cumulative, it's only based on the previous season and that that that change in soil issuance resulted in a dramatic decrease in the in the pod rate is that the padre itself but the change in the pod rate over time and then ultimately it led to the first decrease in the rates of you go to the the charts on being that money you can see I think it

was in I don't have it in front of me right now, but I think it was in like November of 2021 when that bit went live and that that was one of the most significant changes in the model. And originally it was thought that there was some lag. I mean, it was kind of stupid and in hindsight, but the to still me and the argument, the concept was that if there was there was some common ation of time factored into the soil issue.

And so it's not just the magnitude of the current deviation, but it's a time weighted average over a longer, not a time weighted average, a a sum over time. Excuse me, The concept is that that would account for not just the current deviation, but some some some notion of the magnitude of the deviation. But time was clearly, in practice, not the right value to consider in terms of soil assurance.

Now, it may be something to consider on the temperature front, but but it was not a good thing to have on the soil. Issuance from. Okay, so where we are right now, there is soil available. And as you said, maybe the temperature is not high enough or that is not, you know, enough demand for that soil at the moment.

But it's very easy to forget that it was not long time ago when, you know, the stock was selling out or selling soil, but it didn't even need to or necessarily need to. And I think by some accounts, almost half of the debt that's available right now by being slack was issued, you know, due when it was unnecessary or when the price of bean was above one.

And as you said, that was one of the major changes or updates to being sold. But now it only shows the amount of soil that is necessary or that is needed to to get back to Peg, let's talk about another improvement and that that I believe happened after after replant, which is the amount of soil that is now issued, depending on that that level as is different, you know, when the price of beans above one, maybe I'll give an example.

So let's say beans typically issue 100 beans now that the price of beans above one of the that is too high. It would on the issue of 50 beans and that has the effect of reducing the pod line instead of maintaining it 1111 the price of beans above one. And I think the answer to here why and why is because, you know, the debt level is high.

So being so-called optimized was also measuring the still continue to measure the demand for soil but also reducing the the problem is that correct problems. Yeah And at this point there's a a further improvement under under on at the moment to implement a Dutch auction in the field such that the beginning of each season I believe I believe it'll probably be the first 25 seasons, the first 25 blocks excuse me, a season.

There's a Dutch auction such that the temperature is priced up to the current temperature and that will hopefully also dramatically improve the soil issuance. And there's there's again, there's a lot of there's a lot of open questions as to further improvements that can be made to the to the temperature. But as as you said, there was another improvement on the soil issuance and these are all marginal changes.

And there's I mean, there's a couple there's a couple of things that can probably still be improved. Okay. And yes, the Dutch auction is expected. Again, that happens one, when the price is above one and we're selling out of soil. But it's easy to to look at where the cities are now. And for got, you know, the other the other side of the coin I guess when when the price is above one of the challenges I would say in issuing a showing that so the Dutch auction would would have hopefully help with two things.

First often is reduce the amount of that that is a should or optimize on the amount that is sold and also help with measuring or more accurately, the same measure of the the demand for that for that that you touched a little bit about maybe other future improvements. What do you think what what other ideas or things that you think you know are ways to think more about, you know, improving, improving the field?

So again, from a theoretical perspective, in addition to so so the two values that Bienstock currently considers is the the B and the pod rate. There is a question of magnitude of deviations in terms of time and deviation from the peg. And I mean, you can you can perform as much statistical analysis on based on those inputs as much as you want in theory and hopefully at some point people do the, the, the analysis or the theoretical analysis on what is the theoretical optimal for things to measure at the margin.

So those are like the theoretical inputs that are worth considering. And we have not really spent the time on that. Don't intend to. There's too much stuff to to think about and that's really only going to be a marginal improvement and it's unclear if it's if it's something that can be answered any time in the near future, particularly given the small scale of being stock.

So it might just be something that takes a lot of time to get a lot of data on before really those all of those factors can be considered, but something that's a simple improvement that could be made in the near future to the to the field that would likely improve things would be to have some downward reflexivity in the temperature when the system re pegs, perhaps if it repents for a certain period of time, such that in addition to the FIFO incentive, there's also a a larger risk at the at the margin of the interest rate that you'll receive going down in addition to your place in line going up.

So that could potentially create an even more efficient market for soil. But other than that, you know, that's probably I mean, there's also an open question as to whether it makes sense to have a a I mean, one thing we haven't talked about is demand for soil. And I'm sorry for kind of skipping some of the stuff, just have limited time to for class, unfortunately.

But the the measurement of demand for soil is something that is highly imperfect in a lot of different ways, particularly. And it's very hard to measure when there has been demand for all of the soil because as soon as the system defines how it's going to read all of the soil being sown in the season, it the the lenders can can so just the parameter minus a very small amount such that no it's not worth the gas beta.

So to the the parameter that would be considered the maximum soil and in short that makes it hard for the system to know whether in practice all of the soil has been sown. And so there's some open theoretical question as to how to measure demand for soil currently being stock measures, demand for soil in three states. And there's a an increasing steady and decreasing demand for soil.

And there's an open question as to whether there is a need for a steady state. And so in addition to the the delta B and the pod rate in practice, the changes in temperature are a function of the change in demand for soil from season to season. Another thing worth considering is whether or not the change in demand for soil should be based on a time weighted average over multiple seasons.

Currently, it just is a from from the previous season to the current season and whether or not there needs to be a steady state or not. So currently there's 24 different states for temperature changes based on the three parameters. And again, that can can be generalized into infinite more of a continuous state as opposed to discrete 24 discrete states.

But at the moment that's probably another place where the system could be refined and otherwise, you know, that's probably enough potential improvements for the moment or just commentary on where there's inefficiency, even without solutions. Okay. And untrue, as you said, even even though we dedicated this class for the field, it feels like we can throw a lot of all of the details and rightly so.

We can spend a lot of time, maybe the multiple classes just discussing discussing the field. Well that I will I will pause here and see if the audience has any questions that we can answer regarding the topic of today or any any other question that is before before ending this class. Okay. I'd like to remind everyone again that there are two channels of that for discussions of the two bips that are going to be proposed.

The two groups, one is for set. Set off is the third the solution by our auditor that is meant to help secure the second going to be used in governance. And then the other bit is for the business performance budget. Notably again, the difference in this budget is that is going to be for the half of the year.

So it's not a quarter of the budget and encourage everyone again to read through the proposals, join the discussion and then we have a meeting tomorrow that you can come in and, you know, discuss at a glance. The team will also hold office hours that they will be happy to to answer any questions about both these proposals and again, about about this.

And Jared, thank you all for joining us today. And we'll see you next week.