Group

Frequently
Asked Questions

bg

Background

What is Upshot One?

Upshot One is a question and answer protocol. It leverages a new field of mechanism design called peer prediction to incentivize people to answer questions honestly. It’s purpose is to reliably and efficiently pair honest answers to questions. This enables and improves a number of exciting use cases such as efficient price discovery for esoteric assets, decentralized insurance, decentralized governance, content curation, and more.

What is peer prediction?

Peer prediction is a nascent field of mechanism design that explores new ways of eliciting quality answers from participants. Peer prediction accomplishes this by considering the structure of information. More specifically, peer prediction mechanisms use the mutual information (a generalization of correlation) between sets of answers to identify truthfulness of those responses. The more honest a set of answers is, the more mutual information they admit.

Why did we choose to tackle the NFT pricing problem?

NFTs constitute an exciting new asset class whose true potential has yet to be realized. We blame this largely on their lack of efficient price mechanisms, which is an understandably difficult undertaking; NFTs are a type of asset that is both non-fungible and low-velocity (much like art and real estate), so it is difficult to price NFTs from open market transactions alone. However, with such a price mechanism, we believe that new, valuable financial instruments could be made. These instruments may otherwise look familiar but utilize NFTs in ways yet unseen. For example, it would be possible to collateralize loans with digital art in a trustless manner. It would be possible to create exotic indices of NFTs such as indices of interest in video games based on the aggregated valuations of their in-game NFT rewards. The bridge between this reality and the present is a protocol that generates high-quality information in the form of accurate valuations of NFTs, and Upshot One is capable of being that bridge.

Why can Upshot One tackle the NFT pricing problem?

With Upshot One, we can coordinate decentralized communities of users to honestly appraise the value of different NFTs. Upshot One is a question and answer protocol that pairs questions to honest answers. Hence, we can use Upshot One to map “What is the price of this NFT?” to an actual price that people genuinely think is appropriate.

Where is Upshot going?

Upshot One is a modular protocol that can be used across several important verticals; the efficient price discovery of NFTs is just its first implemented use case. Over the next few months, we will be improving our NFT appraisal product as well as building products in insurance, governance, and content curation. We will also be progressively decentralizing its technical stack.

How it Works

What are messages?

Messages are the most fundamental objects passed between users and Upshot One. There are four primary types of messages: questions, answers, resolutions, and scoring rounds. Question messages specify the query being made to the users (e.g. “Which NFT is more valuable?”) and any additional information required (e.g. the possible answers to the question and the latest time at which this question can be resolved). Answer messages are created by individual users that contain their answers to specific questions (e.g. “The NFT A is more valuable than NFT B.”). Resolutions are created as a result of scoring rounds and represent the final answer to a question. For every question, there is at most one resolution. Scoring round messages contain the information that went into a specific scoring round, including the committee (see “What are scoring rounds?”), the answers provided by everyone in that committee, the questions resolved, their resolutions, the random seed used to select the committee, and more.

What are groups?

Groups are how different communities are specified in Upshot One. A group has two primary roles. First: Groups specify the category of question that is to be asked in it. For example, a question which asks users to compare two NFTs is similar to any other NFT comparison question, but not similar to asking users to state their favorite movie. Second: Groups specify the criteria users must satisfy before being allowed to answer questions in a group. In the case of PunkMash, a user must either hold CryptoPunks themselves or UPT tokens.

What are strategies?

Strategies determine the eligibility of users in groups. They read chain data and interpret it in different ways to assign a weight to a user. This weight is used to determine how often a user’s answers are incorporated into a scoring round. For example, the two strategies used by the MEMEMash group are erc721-balance-of and erc20-balance-of, meaning that if a user has a balance of MEMES NFTs or UPT ERC20 tokens, they are eligible to answer questions in MEMEMash. Furthermore, if the user has the most MEME NFTs and the most UPT ERC20 tokens, then their answers have the highest chance of being selected for a scoring round in MEMEMash.

What are scoring rounds?

Scoring rounds are periodic activities that score users’ responses to questions and create resolutions for questions. A group will specify how frequently scoring rounds occur. A scoring round starts by randomly selecting a committee of group members that have recently answered enough questions. The committee size is generally somewhere between 3-5 members and the minimum questions a member must have answered to be eligible is usually 4 and never less (these parameters are set by each group). With a valid committee, sets of answers to both resolved and unresolved questions are chosen for each member. We require that each committee member has answered at least 2C of the same questions, where C is the number of possible answers to each of the questions (this requirement is necessary to achieve the truth-telling guarantees of our peer prediction mechanism of choice). Now that we have the committee and questions that will be scored, we score committee members against one another using the DMI-Mechanism, which uses a novel mutual information measure to score each participant by how honest their answers are. The result of this is a payment owed to each member (which is their truthfulness score) and resolutions to all of the unresolved questions included in the scoring round.

Getting Started

How do I join a group?

There is no concept of “joining” a group. You are either eligible or ineligible to answer questions in a group based on a group’s specific eligibility criteria. If you answer enough questions in a group and if you hold certain tokens (e.g. CryptoPunks NFTs in the PunkMash group), you will be considered for scoring rounds and paid based on how honest your answers are.

How do I answer questions?

Answering questions is as simple as signing a message that contains the necessary information to classify your answer. This includes the ID of the question you are answering and your answer to the question.

What is weight?

Weight is how much influence you have in a group. In other words, how often you are going to be selected for scoring rounds. Your weight is determined by the strategies used in a group. For example, the ArtMash group places 1 unit of influence weight for every NFT you own from SuperRare, KnownOrigin, and MakersPlace. So, if you own 2 NFTs from SuperRare and 1 NFT from KnownOrigin, your weight (i.e. influence) in the group will be 3.

When are my answers used?

Users are selected to be committee members randomly according to their weight; the higher a user’s weight, the higher likelihood that their answers will be included in the next scoring round.

Who pays me?

Capital for payments is sourced from at least two avenues: question rewards and yield. Question rewards are paid by each respective question creator. They are disbursed to committee members in proportion to their truthfulness in the successful scoring round that resolves that question. Users may stake in groups to foster the pairing of answers to questions they care about (e.g. an NFT community may be interested in learning the value of the NFTs they own). Staking creates “fuel” for groups in the form of yield. When someone stakes in a group, their staked capital is invested in different interest-earning protocols, on-chain. The yield earned from these protocols is divided among users included in scoring rounds proportional to their calculated truthfulness in those rounds.

What am I paid in?

At first, all users are paid in UPT, but this will vary with time. You can be paid in any currency, and this can be determined by groups or by question creators. Groups can specify a payout currency of their choosing or defer this decision to question creators.

Can I spend UPT?

No. UPT is moreso for tracking influence in a given group and for enabling the use of additional incentives (e.g. selection bonuses). It is a non-transferable token on Ethereum.

Understanding Metrics

What is a selection bonus?

A selection bonus is an additional, optional incentive groups can put in place to incentivize participation by its members. If a user answers a certain number of questions in a specified period of time, that user can be given additional UPT tokens to increase their influence in the group. These bonuses are often tracked and distributed on a weekly basis. For example, if you answer 100 questions in a week in ArtMash, MEMEMash, or PunkMash, you are given an additional 30% of UPT tokens at week-end to increase your chances of being selected for scoring rounds in all subsequent weeks.

How is my truthfulness calculated?

Your truthfulness score is calculated using the DMI-Mechanism, which is run during scoring rounds. It scores you based on the amount of mutual information your answers share with other committee members’ answers. The way to maximize this score is to be informed and honest. In DMI-Mechanism, there is no strategy of answering questions that is more profitable than expending effort to answer them honestly. The truthfulness score in your rankings for a group displays the difference between your truthfulness score and the maximum truthfulness score you could’ve gotten.

What does my expected frequency of being selected for scoring rounds mean for how much I earn?

You get paid whenever you are selected for a scoring round. This means that the more scoring rounds you are selected for, the more you will get paid. You are selected for scoring rounds based on how much influence you have in a group.

Pricing Assets

How are everyone’s answers aggregated?

Everyone’s answers exist as messages on Skynet. When a committee is formed for a scoring round, the committee members’ answers are read from Skynet and a subset of them is chosen to be used in the scoring round. Some of these answers are for questions that have already been resolved, and some are for questions that are unresolved.

How do we turn NFT comparisons into predicted prices?

Resolutions to NFT comparison questions tell us the relative value users place in one NFT over another. However, they do not give us a dollar-value for the compared NFTs. To turn these relative rankings into prices, we first analyze the open-market prices of other, higher-velocity NFTs. Next, we ensure that the pairs of NFTs compared always include one higher-velocity NFT and one NFT whose price is uncertain. Thus, resolutions to comparison questions yield concrete bounds on the real value of NFTs. For example, if we have market data on NFT A (a higher-velocity NFT, that, by definition, has been transacted) and know that its price is about X (since it’s been transacted, we at least can estimate its current value) but none on NFT B (whose price we’d like to know), we may compare NFTs A and B. If A > B, then we know the value of B is less than X. By repeating this process over many comparisons, we can achieve a tight interval on the value of NFT B.

Why don’t you just directly ask users for the price of NFTs?

This is possible to do with Upshot One but it’s tricker to implement in a user-friendly way. That being said, we will be releasing groups with this feature in the future.

Upshot aims to provide real-time price feeds for NFTs through crowdsourced appraisals – empowering a new generation of appraisers and catalyzing powerful new innovations at the intersection of DeFi x NFTs.