bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 3쪽 중 1쪽
Tellor
Last Edited: 11 Aug 2026
Table of Contents
Introduction ............................................................................................................ 2
How the Chain Works .............................................................................................. 2
Submitting Values ............................................................................................................ 2
Cycle List ......................................................................................................................... 3
Disputing ......................................................................................................................... 4
Validator penalties ........................................................................................................... 6
Tipping ............................................................................................................................ 6
Validator Set Dynamics .................................................................................................... 6
Dual Delegation - Reporting vs Validating .......................................................................... 7
Tokenomics ............................................................................................................. 7
Bridging .................................................................................................................. 8
Updating the Validator Set ................................................................................................ 8
Tellor Token Bridge ........................................................................................................... 8
Data Bridge ...................................................................................................................... 9
Signing Prices ..................................................................................................................................... 9
Relayer .............................................................................................................................................. 9
Relay fallback ..................................................................................................................................... 9
Other cross-chain methods .............................................................................................................. 10
Data Usage ............................................................................................................ 10
Robust Data Usage (aka Consensus data) ......................................................................... 11
Edge Data Usage (aka Optimistic data) ............................................................................. 11
Additional data security .......................................................................................... 11
Dispute Monitoring .......................................................................................................... 11
Optimistic as Fallback ..................................................................................................... 11
Additional Self-Driven Security and Fallbacks .................................................................. 12
Fork Choice ........................................................................................................... 14
Soft Upgrades ................................................................................................................. 14
Hard Forks ...................................................................................................................... 14
Plan for legacy Tellor contracts ............................................................................... 14
Conclusion ............................................................................................................ 14
Glossary ......................................................................................................................... 15
Introduction
From the initial deployment in August 2019, Tellor has been designed so that no trusted party can
prevent users from requesting, reporting or disputing data. Today we continue with the same
principles: permissionless validators, reporters, disputers, relayers and users. Anyone can become
a validator or reporter via staking, dispute for a fee, get any data by tipping it, and relay the openly
available data attestations on Tellor, without the team’s intervention. Crypto economic incentives
have been in place since inception as we have focused our eEorts on being decentralized and
secure. Tellor ’s design supports data liveness guarantee which is only possible because the design
allows the protocol to survive even if the team disappears1.
Every aspect of Tellor is open source including consensus, bridging, reporting, and relaying so that
everyone can audit and participate in the network as validators, reporters, users, and holders.
Throughout the protocol, architecture decisions prioritize censorship resistance, open
participation, and security over speed. Slow unstaking, delayed withdrawals, dispute windows,
dispute rounds, configurable user by user trust assumptions such as fallbacks, consumer side
validation and choice of relay all aim to reduce systemic risk.
Tellor provides a flexible foundation for users to choose how they consume our data. Users can
control how much validation, delay, monitoring or redundancy they require and even layer their own
governance or additional security.
How the Chain Works
Tellor is a stand alone L1 built for the purpose of coming to consensus on any subjective data. It
works by using a network of staked parties who are crypto-economically incentivized to honestly
report requested data.
Submitting Values
Any staked reporter can submit a value for a given query (i.e. data request, e.g. BTC/USD price). All
queries have a report time frame. For that given time period, reporters can add their value to the
array of submissions. At the end of the time period, the oEicial value is determined by a weighted
aggregation of all the values submitted.
All queries have an aggregation type associated with it (e.g. median, mode, average, etc.)2. Once
the time frame ends, the reports are subject to weighted aggregation. For this reason, larger
reporters contribute more to the oEicial value than smaller ones. Each oEicial value takes a
minimum of three blocks (target time 2s per block). The first block is the tip (e.g. “I request
BTC/USD”). The second block is the reporting and the third is more reporting and aggregation
phase, where all values are submitted and then aggregated, at which point an oEicial value is
determined for the query. Tips are distributed to all reporters for a given query using their weighted
contribution.
1 This is an ultimate test of decentralization. But team and community contributors help ensure protocols evolve with the
times and improve as new technology becomes available.
2 For prices and most numeric data feeds the aggregation method is the median this makes it more di=icult and costly to
manipulate and this can be customizable per users’ specification. Additionally, o= chain computation of more robust
methods such as TWAPs or VWAPS can be done before submitting to aggregate on chain for the o=icial value.
Figure A: Tellor process from Tip to Data Aggregation.
The reporting time frame begins once a queryId is tipped. All reporters can add their value to the
submission array for inclusion in the aggregation and for distribution of the rewards.3 Tips received
within the time frame are added to the initial tip. If no reports are submitted during the time frame,
the time frame is restarted upon the next tip and the original tip is added to the new tip for that
queryId.
Cycle List
In order to maintain freshness, the system maintains a list of enshrined queries, a cycle list.
Reporters continuously report for the next query in the list to ensure that tips do not need to be
submitted in previous blocks to have a base level of reporting (thus avoiding wasting gas by
validators who just want inflationary rewards). This signals reporters what to report so they act in
unison and contribute as much aggregate power to a single report as is available. Reporters are
never required to report for any queries including those in this list, but as long as inflationary
rewards are enough to cover the gas costs of reporting, some query from the cycle list should be
reported to consensus every block. Cycle list changes are voted on by governance.
To incentivize consistent reporting across all cycle list queries, TBR distribution uses a per-
aggregate power share approach. Instead of distributing TBR immediately per aggregate, rewards
are accumulated over a configurable period and distributed based on each reporter's power share
within each aggregate.
3 The reason for this is that some queries are not automatic, e.g. “let’s type in an answer manually” , so we want to give
room for non-time sensitive queries to get more reports. Spot prices and automated queries will have a report time frame
of 1 or 2 blocks.
Disputing
There are some diEerences in disputes from legacy Tellor but what has not changed is that the
entire dispute and resolution mechanism is done on chain with transparent slashing and with
earning successful disputers earning the majority of the slash amount. This mechanism continues
to incentivize everyone to run a disputable values monitor and to compete to be the first to dispute.
Also, the user does not have to wait for resolution, instead if the value was the median the value is
flagged and removed, and they can use the next available value. Similar to the old Tellor system
too, is the idea that since you can censor a party by disputing, disputes can be seen as a wager on
who is correct (the reporter or the accuser).
What is diEerent is that reporter stakes are not tied to specific values, but rather to a given reporter.
For instance, a reporter can submit ETH/USD once every second with their same stake. If they
submitted a bad value 2 days ago4, they can still be subject to a slashing event. Any party can raise
a dispute with free floating TRB, but unlike the old system, reporters and validators can use their
stake (or part of it) to begin a dispute. Once initiated, the dispute fee and the potential slash
amount (from the accused reporter) are put in escrow and removed from corresponding staking
powers.
To initiate a dispute, the disputing party submits a dispute against a given reporter for one of three
categories:
- Warning (dispute fee is 1% of stake) - jail5 with no minimum time lock, can call a
function to be released from jail and begin reporting again
- Minor Infraction (dispute fee is 5% of stake) - jailed for 10 minutes and out when
they call the release from jail function
- Major Infraction (dispute fee is 100% of stake) - jail until dispute over (since 100% of
stake).
A release function has to be called after a warning or minor infraction to ensure the staker has
looked at the dispute and implemented a fix as necessary. Infractions in these lower two tiers can
generally be assumed to not be malicious.
After specifying the dispute category, the disputer submits an amount of TRB up to the minimum
slashing amount before the dispute can initiate. If they don’t have enough funds themselves, for up
to one day, others can add to the pot until they hit the slashing amount (i.e. 1, 5, or 100 percent
depending on the slashing category). Once the amount is hit (could be hit instantly upon proposing
the dispute, or could take up to a day), the potential slashing amount will be taken from the
disputed validator and placed into a locked stake.
4 Because of the competitive forces in Tellor, validators, reporters, users, and disputable value monitors(many of which
are other validators and reporters) a reporter and bad value would be disputed almost immediately. Users and monitors
can observe data real time on-chain for Tellor and dispute before it has the chance to get to their mempool and be subject
to MEV and OEV . Transparency helps prevent our users from ingesting bad data because it does not wait until after the
fact for disputes and slashing and has direct bounty to incentivize watchdogs/monitors.
5 “Jail” is a concept in the tendermint system where the validator (or reporter in this case) is locked out of participating for
a certain period of time
For up to two days, stakeholders will vote on which side of the dispute they support. Tellor uses a
diEerent voting distribution6 from the legacy Tellor governance. Previously, voting power was evenly
distributed among users, reporters, token holders and the team. However, because most TRB on
Tellor is expected to be either staked or used for tips, the voting distribution is split among these
three groups:
33.3% users (tips)
33.3% reporters
33.3% team
When reporters vote they vote with the backing of all the TRB that was delegated to them. However,
the selectors (TRB holders that selected a reporter) can independently vote for the dispute and
overwrite the reporter’s vote for their voting power. Holders can help increase the security of Tellor
by delegating to validators and selecting a reporter as this will allow them to participate on dispute
votes.
Once the two days are over there is a one day period where the dispute can be reopened and the
same two day voting round is repeated. However, if at any point a quorum of >50% total voting
power votes in favor of one side of the dispute, the dispute is considered finalized and no new
rounds can be opened.
Once the dispute is resolved, the stake from the losing party is transferred to the winning party(ies)
as undelegated, staked TRB. Tokens disputed or used as fees in disputes are not released as free
floating tokens so as disputes cannot be used as a way to exit staking faster, however if free floating
tokens are used to pay the fee, they are returned as free floating tokens in the case of a successful
dispute.
Note that each dispute round (even the initial one) takes 5% of the dispute fee. Of that 5%, half is
burned and the other half is divided amongst voters in the system. The dispute fee then doubles
each round up to the slash amount to further incentivize voting and to prevent spamming. Once
quorum is reached on a dispute, further dispute rounds cannot be raised.
For usage purposes, values are not attested to if the disputed reporter was the oEicial aggregated
value (median contribution). Values that need to be flagged are also added to evidence when the
dispute is initiated (an array of values). For this reason, users who rely on optimistic finality should
be aware of dispute censoring attacks and the potential for values to take time to get through.
Tellor ’s on chain dispute mechanism, resolution and slashing allow it to be a preventive measure as
well as a proactive one. Slashing helps deter attackers while also economically incentivizes
anyone, including the user to be proactive in disputing before the data is ingested by a protocol.
After the fact disputes are also allowed, however, the incentivized preventive and proactive
approaches are preferred and properly incentivized since once the data is ingested, damage could
be irreversible.
6 Checkout our ADR 1008 on voting power by group for more information on why validators are not part of it
https://github.com/tellor-io/layer/blob/main/adr/adr1008%20-%20voting%20power%20by%20group.md
Validator penalties
Validators have normal cometBFT liveness and double-signing penalties, which are handled
automatically by the chain. Additionally, Tellor applies slashing for malicious attestations and
malicious validator set update signatures. This is done by manually submitting evidence of these
behaviors.
Standard cometBFT penalties Additional Tellor penalties
Liveness (15 minutes of inactivity) slash is 1%
Malicious attestations are slashed 1% of stake
Double-sign slash is 5% Malicious validator set update attempts are
slashed 5%
Tipping
All tips are done in TRB. Each query can be tipped directly and its tip will increase as more users tip
it7. Once a report is submitted and aggregated, 98% of the tip is split amongst reporters for that
query in that given block. There is a 2% fee on the tips (to prevent vote farming/ spamming). This 2%
fee is burned. Tips are distributed as locked TRB. Once locked, parties can run functions to claim
the tip, in which the tips are unlocked and added to their stake or to tip balance
(‘MsgWithdrawTipToBalance’) which will start the clock to unbound these in 21 days. This is to
ensure that parties cannot bypass validator deposit limits through tipping, as well as to prevent
farming vote power via the tipping mechanism.
Note that tipping consists of only one time tips. There is no built-in heartbeats or price thresholds
because the complexity added is unnecessary and better handled oE-chain (e.g. AI agents, or tip
bots like the autoTipper and relayer handle this functionality already).
Validator Set Dynamics
Chains using the tendermint consensus mechanism have a limited number of validators. By setting
the number of validators, this allows for eEicient interconnectedness between chains as well as
faster throughput. The tradeoE that any chain must consider is that more validators leads to slower
blocktimes and more expensive bridging costs. Tellor has started with a limit of 100 validators but
will move to a larger set as technological advances or market conditions allow (e.g. if no one needs
sub-2 second blocks).
The validator set can only change by a maximum of 5% per 12 hours including tip claims. This is for
purposes of maintaining a stable validator set for bridging eEiciency. Once the 5% change is hit,
new validators will need to wait until the rolling percent change per 12 hours is under the cap. This
5% is a parameter that can be changed via governance.8
7 Users that need the same data feeds can crowdsource tips. On-chain tips make costs transparent and AI agents can
interact with Tellor easily as they can request support for new data without needing humans to sign a contract.
8 There is also a delay (21 days) to exit (the unbonding period).
Dual Delegation - Reporting vs Validating
Tendermint uses a delegated proof-of-stake(dPoS) model where there is a set number of validators,
but all token holders can delegate to validators to increase security and share in rewards. Tellor
uses this delegation but adds a second delegation for reporting duties, which we refer to as
selecting. Each token can be used as a stake for reporting and for validating. Parties can delegate
both the reporting and validating to the same party, to diEerent parties, or even to themselves. The
same token is subject to slashing by either method (reporting data or failing to honestly validate the
chain) and the stake balance for both delegations is reduced immediately upon either
consequence, so great caution should be taken when delegating and/or selecting.
Figure B: Dual delegation
Note: The delegated validator and the selected reporter can be di5erent(A), the same(B), mix of yourself
and others(C).
The reason for this dual delegation is that validator sets are capped in tendermint based systems,
however we need to remove that cap to enable smaller and more reporters to help decentralize the
data provider set. Additionally, the cost of bridging is directly tied to the validator set size (verifying
signatures for the light client bridges), so a large validator set such as Ethereum is unfeasible for
our intended uses (the need for fast, cheap bridging of data).
Tokenomics
TRB is the native currency of the chain프로젝트로 돌아가기