[Constitutional] AIP Fast Feed

Fast Feed: Constitutional AIP

Abstract

This Constitutional AIP proposes to establish the Fast Feed, a paid, authenticated data streaming product for Arbitrum One. The Fast Feed allows subscribers to get updates on transactions, their relative ordering, and related transaction metadata earlier than they would otherwise via the regular feed. The Fast Feed makes that stream accessible to Arbitrum’s ecosystem as a specialized, subscription-based data product.

This feed will be paid-only to mitigate against DoS risks and to help ensure a reliable and stable connection for teams who wish to use it. The Fast Feed’s design should not introduce new types of MEV, frontrunning, nor sandwich attacks because the feed only publishes read-only information about the transactions after their order has been determined by the sequencer and enqueued for execution.

If approved, this AIP would:

  • Be paired with an Arbitrum Nitro node software upgrade at the sequencer to expose a new endpoint for access to the Fast Feed.
  • Deploy and use the Fast Feed Payment Contract, an audited smart contract on Arbitrum One, to manage access to the Fast Feed by allowing users to submit their API key hash when they send a paying transaction to the Fast Feed Contract. Upon payment, the Fast Feed Payment Contract records the hash of the key as the user’s Subscription Credential, which the sequencer cross-references to admit or deny incoming Fast Feed connections.
    • The Fast Feed Payment Contract will use a RewardDistributor contract as a recipient of fees such that 97% of subscription revenue gets routed to the ArbitrumDAO and 3% to the Arbitrum Developer Guild. The RewardDistributor contract has been audited and a public audit report can be found here.
  • Grant rights to Offchain to make adjustments from time to time for a period of two (2) years to the parameters in the Fast Feed Payment Contract.

Motivation

The Fast Feed is designed to serve any ecosystem participant that benefits from earlier visibility into ordered transaction flow (including MEV searchers, Proprietary Automated Market Makers (propAMMs) and latency-sensitive DeFi applications). Access to the Fast Feed is provided on a paid subscription basis, with the goal of using the proceeds to help ensure quality and consistent feed performance for all subscribers. The net revenue will be split between the ArbitrumDAO Treasury and the Arbitrum Developer Guild pursuant to the RewardDistributor contract deployed on-chain, following the same revenue split framework established by the Timeboost AIP.

Rationale

The Fast Feed distributes information from the sequencer before a full block is built, but because it is optional, supplemental, and costly to provide, it will be a paid product. Once ordering is resolved deterministically, that ordered stream carries clear economic value for latency-sensitive participants. Rather than letting this value accrue only to those who can co-locate or operate the fastest infrastructure, the Fast Feed packages it as a subscription product whose proceeds flow back to the ArbitrumDAO and the Arbitrum Developer Guild.

  • Open, not exclusive: access is paid but permissionless; any participant willing to subscribe can obtain access, and there is no privileged or preferential access beyond payment.
  • Ordering-neutral: the feed does not change transaction ordering, inclusion guarantees, or the priority or transaction fees any user pays. It only changes the timing by which subscribers are able to observe the ordered stream of transactions.
  • Transparent, on-chain revenue: proceeds are split through the RewardDistributor and enforced on-chain (97% to the ArbitrumDAO Treasury, 3% to the Arbitrum Developer Guild), with every payment and distribution emitted as an on-chain event for public audibility.
  • Consistent governance precedent: revenue routing follows the framework established by Timeboost and the proposed transition to use PGA, keeping Arbitrum’s ordering-related revenue products aligned.
  • The ArbitrumDAO retains control: the ArbitrumDAO and, if the DAO permits, Offchain can configure the new feed payment parameters (see section 1.3). The DAO alone can additionally control the revenue split.

Key Terms

Fast Feed: an authenticated, low-latency data stream that publishes the individual transactions, their relative ordering, and relevant metadata after they are ordered by the sequencer, available to paid subscribers.

Sequencer Feed Endpoint: the Fast Feed endpoint exposed by the sequencer by which authenticated subscribers can connect to for access to the Fast Feed.

Fast Feed Payment Contract: the audited smart contract on Arbitrum One that manages access to the Fast Feed.

Subscription Credential: the API key hash recorded by the Fast Feed Payment Contract upon payment, which the sequencer cross-references to admit or deny incoming feed connections.

propAMM (Proprietary Automated Market Maker): an automated market maker whose pricing relies on an external trusted price-pusher source to update quotes on-chain rather than relying solely on the constant-product curve and arbitrageurs. propAMMs are an emerging class of actors who offer attractive swap prices/quotes and usually bring their own liquidity on-chain to facilitate this.


Specification

We formally introduce the Fast Feed which consists of three components: 1) the Sequencer Feed Endpoint and payload, 2) the Subscription and Credential Management, and 3) the Payment Contract itself.

1.1 Sequencer Feed Endpoint and Payload Schema

A new WebSocket endpoint that publishes individual transactions once they are ordered but before a full block is constructed and emitted from the regular feed. The endpoint is authenticated: only clients presenting a valid subscription credential are admitted. Note that the Fast Feed will only publish information about transactions after these transactions have been sequenced for execution. The ordering cannot be influenced or changed past this point and, therefore, types of MEV like sandwiching or frontrunning should not be able to be performed.

Each transaction is wrapped in a FeedMessage envelope that provides ordering and timing metadata.

{
  "FeedMessage": {
    "description": "Envelope wrapping each transaction with ordering and timing metadata.",
    "fields": {
      "version": {
        "description": "Protocol version for forward compatibility",
        "type": "uint32",
        "json_type": "number"
      },
      "timestamp_ms": {
        "description": "Unix timestamp in milliseconds when the sequencer created this message",
        "type": "uint64",
        "json_type": "number"
      },
      "pga_round": {
        "description": "Number of the PGA round for the current block",
        "type": "uint64",
        "json_type": "number"
      },
      "transaction": {
        "description": "Transaction data and incomplete receipt",
        "type": "IncludedTransaction",
        "json_type": "object",
        "fields": {
          "block_number": {
            "description": "Tentative block number (not guaranteed to be finalized)",
            "type": "uint64",
            "json_type": "number"
          },
          "tx_index": {
            "description": "Transaction position within the block so far",
            "type": "uint32",
            "json_type": "number"
          },
          "raw_tx": {
            "description": "RLP-encoded signed transaction",
            "type": "bytes",
            "json_type": "string (hex-encoded)"
          },
          "tx_hash": {
            "description": "Transaction hash",
            "type": "bytes32",
            "json_type": "string (hex-encoded)"
          },
          "receipt": {
            "description": "Execution result. Incomplete because the block is not yet finalized; does not include block_hash.",
            "type": "IncompleteReceipt",
            "json_type": "object",
            "fields": {
              "status": {
                "description": "1 = success, 0 = failure",
                "type": "uint8",
                "json_type": "number"
              },
              "gas_used": {
                "description": "Gas consumed by this transaction",
                "type": "uint64",
                "json_type": "number"
              },
              "gas_used_for_l1": {
                "description": "Gas attributable to L1 data costs (Arbitrum-specific)",
                "type": "uint64",
                "json_type": "number"
              },
              "cumulative_gas_used": {
                "description": "Cumulative gas used in the block up to this transaction",
                "type": "uint64",
                "json_type": "number"
              },
              "effective_gas_price": {
                "description": "Actual gas price paid (relevant for PGA priority fee analysis)",
                "type": "uint256",
                "json_type": "string (hex-encoded)"
              },
              "base_fee": {
                "description": "Current block base fee",
                "type": "uint256",
                "json_type": "string (hex-encoded)"
              },
              "contract_address": {
                "description": "Set only if the transaction created a contract",
                "type": "address (optional)",
                "json_type": "string (hex-encoded)"
              },
              "logs": {
                "description": "Event logs emitted during execution",
                "type": "[]Log",
                "json_type": "array of objects",
                "array_item_fields": {
                  "address": {
                    "description": "Emitting contract",
                    "type": "address",
                    "json_type": "string (hex-encoded)"
                  },
                  "topics": {
                    "description": "Indexed event fields",
                    "type": "[]bytes32",
                    "json_type": "array of strings (hex-encoded)"
                  },
                  "data": {
                    "description": "Non-indexed event data",
                    "type": "bytes",
                    "json_type": "string (hex-encoded)"
                  }
                }
              }
            }
          }
        }
      }
    }
  }
}

1.2 Subscription and Credential Management

The Fast Feed Payment Contract governs subscription validity. Access to the Fast Feed is granted when a user submits a paying transaction to the Fast Feed Payment Contract that includes the Keccak-256 hash of their API key. Upon payment, the contract records this hash as the user’s Subscription Credential. To connect, the user presents their raw API key to the sequencer, which hashes it with Keccak-256 and cross-references the result against the recorded credentials to admit or deny the connection.

Credential rotation and subscription renewals are handled through direct interaction with the Fast Feed Payment Contract by the subscriber.

1.3 Fast Feed Payment Contract

The Fast Feed Payment Contract is an audited, smart contract deployed on Arbitrum One. This contract will:

  • Sell access to the feed with pricing determined using the same pricing mechanism in EIP-4844; market parameters are listed further below.
  • Receive payments denominated in an ERC20, USD-backed stablecoin. We explicitly leave the asset selection to the Arbitrum Treasury Management (ATM) Council and OAT to ensure alignment with the portfolio’s goals.
  • Emit on-chain events for each subscription payment and distribution, enabling transparent public audibility of all revenue flows.
  • Have 3 roles built-in:
    • DEFAULT_ADMIN → controlled by the ArbitrumDAO and has admin permissions to assign all other roles to accounts.
    • BENEFICIARY_SETTER → controlled by the ArbitrumDAO and has permissions to change the destination of the proceeds. This AIP proposes that this be set to the RewardDistributor contract address.
    • MARKET_PARAMS_SETTER → proposed to be controlled by Offchain and the ArbitrumDAO such that either entity can change the parameters in Table 1 below.
  • Send proceeds to the RewardDistributor contract, which is in charge of routing 97% of proceeds to the ArbitrumDAO and the remaining 3% to the Arbitrum Developer Guild. The RewardDistributor contract has been audited and a public audit report can be found here.
    • The RewardDistributor contract is deployed at 0xa4a8a4e2fe847af59d340926adcdef6e988bb1f1 and will route proceeds to the following addresses, with the above specified weights:
      • The ArbitrumDAO L2 Treasury Timelock (97% weight): 0xbFc1FECa8B09A5c5D3EFfE7429eBE24b9c09EF58
      • The Arbitrum Developer Guild (3% weight): 0x6A075E9a02eef6978DD66cB63DE430a8c0C419E9

The contract will be independently audited by Trail of Bits prior to deployment, with the audit report published publicly soon after (but before an eventual onchain vote). All contract addresses will be documented in the Arbitrum One developer documentation.

Should this AIP pass, the DAO grants Offchain the permission to change the Fast Feed Payment parameters listed below in Table 1. Note that the ArbitrumDAO may also change any of these parameters at any time via the MARKET_PARAMS_SETTER role in the Fast Feed Payment Contract described above.

Table 1: Fast Feed Payment contract parameters

Parameter name Description Proposed default value
Round duration (in hours) Length of each sales round, the recurring period over which a fixed allotment of feed tickets is sold at a single uniform price before the mechanism re-prices and the next round opens. Also sets the cadence at which newly purchased tickets become active and prior-round tickets expire. 24
Target tickets per-round The per-round sales volume the pricing curve targets, analogous to EIP-4844’s target blob count. 100
Maximum tickets per round Hard limit on feed tickets sold in any single round. 200
Minimum price (in USD) The price floor for a ticket. 17
Price update fraction* Sensitivity of the EIP-4844-style pricing curve, analogous to the blob base-fee update fraction. 144
Excess tickets sold The running tally of tickets sold above target across all rounds that drives the pricing curve. Exposed as a settable parameter so the curve can be re-anchored to a chosen starting price when the floor or sensitivity is changed. 0
Grandfather period fraction The portion at the start of each round reserved for renewals, during which only prior-round subscribers may repurchase, up to the number of tickets they held the previous round, before sales open to everyone. 128/256 (or ~0.5)

This yields a maximum next round price of ~2p and a minimum next round price of ~0.5p where p is price, derived by e^((max-target)/144) and e^((0-target)/144), respectively.


Steps to Implement

  1. A post to the ArbitrumDAO Forums (this post) to initiate productive, collaborative discussions with the DAO and its members, followed by governance call(s) to answer any questions.
  2. Temperature Check (Snapshot): This forum post will advance to a temperature check vote on Snapshot, targeting approximately 1-2 weeks after the initial Forum Post. The temperature check will gauge DAO sentiment on community support for the Fast Feed concept, pricing structure, and revenue split.
  3. Smart contract audit: The Fast Feed Payment Contract will be independently audited by Trail of Bits prior to the on-chain Alt Arbitrum Gov vote, and we will publish an audit report publicly here. We are targeting an audit completion date sometime in July 2026.
  4. Onchain vote (Alt Arbitrum Gov): Following a successful Snapshot temperature check, publication of the public audit report, and final testing: this proposal will proceed to an on-chain constitutional vote for ~14 days. The ArbitrumDAO will be notified on this thread and via official channels before the vote goes live. We intend for this onchain vote to take place sometime in July 2026, barring any delays.
  5. Feed Activation: As soon as the same day the vote is passed, the Fast Feed Payment Contract will be deployed and the Feed will begin broadcasting.
4 Likes

Fast Feed is a smart, well-designed way to monetize sequencer data advantages while directing nearly all proceeds back to the DAO Treasury. It’s ordering-neutral by design, paid to prevent abuse, and follows the successful Timeboost revenue precedent.

Giving latency-sensitive participants (especially propAMMs and sophisticated DeFi apps) a legitimate subscription path instead of forcing them into colocation arms races is excellent for the ecosystem.

Is the subscription pricing model already decided, or will the community/ArbitrumDAO have input on price tiers and payment structure?

With 97% going to the ArbitrumDAO Treasury and 3% to the Developer Guild, does this split feel fair? Should it be revisited periodically?

How will subscription revenue be tracked and reported publicly on-chain beyond the RewardDistributor contract?

Even though it claims to be ordering-neutral, could widespread adoption of Fast Feed create any second-order effects on network dynamics or centralization pressures?

1 Like

Following the forum discussion, the Fast Feed Constitutional AIP has now advanced to the Temperature Check stage on Snapshot here -Snapshot

Supportive of DAO revenue in principle, but I worry Fast Feed creates a paid informational advantage for sophisticated actors while ordinary users get no direct benefit. Since this proposal explicitly offers earlier ordered transaction data to paying subscribers and fits into a broader ordering-revenue framework, the DAO should add public-good earmarks, retail trust safeguards, periodic reviews, and concentration reporting before moving forward @offchainlabs @Arb_Junior

1 Like

Voting FOR. This is the DAO selling a product and earning revenue. That’s the direction I want Arbitrum going, standing strong on its own.

I’ll leave the deep MEV mechanics to the specialists. What matters to me is the structure, a Trail of Bits audit, the same Timeboost revenue split we already approved, and access that’s paid but open to anyone.

Incentives pointed at the token really appreciate this one!

I support the overall direction of this proposal. Turning low-latency transaction data into a transparent, subscription-based service allows the DAO to capture value that would otherwise primarily benefit a limited set of sophisticated infrastructure operators.

I particularly appreciate that the proposal keeps transaction ordering unchanged and routes revenues on-chain with a transparent distribution model. This aligns protocol growth with sustainable DAO funding.

As the Fast Feed is deployed, I believe it will be important for the DAO to periodically review adoption, pricing, revenue generation, and any unintended market effects to ensure the service continues to promote fairness, transparency, and long-term ecosystem health.

@MconnectDAO I appreciate you raising these valid concerns — they align closely with several points I made in my earlier comment that unfortunately went unanswered by @offchainlabs.

While I’m generally supportive of Fast Feed as a smart, ordering-neutral way to generate sustainable DAO revenue (especially following the successful Timeboost precedent), I share your worry about creating a two-tiered information environment. Sophisticated actors and propAMMs getting earlier ordered tx data via subscription could widen the gap for regular users if not carefully managed.

Strongly agree we should bake in: Clear public-good earmarks from the revenue, Retail trust safeguards, Periodic reviews of the model (including the 97/3 split), Proper on-chain transparency/reporting beyond just the RewardDistributor.

It would be great to see these incorporated or at least explicitly addressed, Curious to hear @offchainlabs thoughts on your suggestions.

1 Like

We are tracking the comments and will publish responses soon. Thank you for engaging with this AIP.

Cornell Blockchain is voting FOR this proposal for the following reasons:

  • Gaining even a microsecond edge historically required huge capital expenditures to build custom, optimized node configurations and infrastructure. Legitimizing the Fast Feed as an official, open product shifts the competition away from expensive engineering setups to a standardized, permissionless API. Anyone paying the uniform ticket price gets the exact same low-latency data stream, democratizing access.

  • By giving propAMMs and liquidity providers earlier visibility into ordered transaction flow, the feed allows them to update their quotes before getting caught on the wrong side of a price shift. Mitigating the risk of stale quotes allows market makers to provide deeper liquidity and tighter spreads across Arbitrum, which directly translates to lower execution costs for everyday retail traders.

  • Directing 97% of subscription revenues back to the Arbitrum DAO Treasury–not leaking it to 3rd party investors– establishes a sustainable, non-inflationary revenue stream that supports the ecosystem without diluting token holders.

However, while the feed is permissionless to buy, it is financially exclusionary. If the bidding mechanism drives up the price substantially, small/medium developers will persistently operate in a slower, informationally damaged manner. To determine the impact , it would be interesting to monitor price spikes, as well as if everyday users are facing better or worse slippage/execution prices. This could show if the feed is being used defensively, or if it is being used by predatory bots to optimize execution at the expense of retail fills.

Voting FOR.

Gotta do something if we are retiring time boost… Hopefully this will get some customers. Excited to see it implemented compare the earnings.

What stops a whale from buying up all tickets and essentially blocking access to the fast feed for other actors?

Voting FOR.

Fast Feed is a sound way for the DAO to capture value that today mostly accrues to actors who can afford co-location and custom infra. Packaging that latency edge as a flat, permissionless subscription — with 97% of revenue routed on-chain to the Treasury under the same split we already approved for Timeboost, and a Trail of Bits–audited payment contract — is exactly the kind of self-sustaining revenue direction I want to see Arbitrum take.

The design also holds up technically: because the feed is read-only after the sequencer orders transactions, it doesn’t introduce reordering, sandwiching, or frontrunning risk.

@Arb_Junior - Thank you for the thorough questions. Here are the answers:

  • The subscription pricing model is decided, with connection fees following the same dynamic, demand-responsive model as EIP 4844’s blob fee market.This choice is deliberate as it helps prevent denial-of-service (DoS) attacks, as the price per connection adjusts automatically with usage and rises as more connections are sold. A party trying to buy up every connection to lock others out (DoS) would face a rapidly escalating, prohibitive cost, which makes monopolizing the feed long term economically irrational.

  • The fee split follows the already ratified fee split model used for Timeboost, with 97% going to the DAO and 3% going to the Developer Guild. This revenue distribution is consistent with the Timeboost Fee Split AIP and the proposed move to Priority Gas Auctions AIP. We chose to mirror the existing, community-vetted precedent rather than invent a new split, specifically so Fast Feed stays consistent with how Arbitrum already handles ordering-related revenue. Like any governance parameter, it can be revisited by the DAO if the community decides it should change.

  • All subscription revenue flows through the RewardDistributor contract, so the on-chain record is the authoritative, permissionlessly verifiable source of truth. Beyond the raw contract, these figures can be surfaced the same way Timeboost revenue already is: through public dashboards (Dune, Blockworks) and the DAO’s regular transparency reporting.

  • The Fast Feed is an independent read layer on top of the sequencer: it delivers data faster but gives subscribers no influence over transaction ordering, so it does not create the sequencing privilege that drives centralization in priority-auction designs. If anything, an open, priced subscription reduces centralization pressure by replacing the colocation and hardware arms race with permissionless paid access that any participant can buy.


@MconnectDAO - The fast feed should not expose regular users to malicious activity from third parties. It’s designed to serve both sophisticated users and applications, but it doesn’t create an attack vector against ordinary users. Access is open to anyone; there’s no barrier to acquiring a connection other than the feed access cost itself. We’ll review the model periodically and provide on-chain transparency.

3 Likes

Voted FOR;

This proposal adds a useful feature without changing how the protocol works or how transactions are ordered. Having an official layer on top of the existing infrastructure for “sophisticated users” to access this information feels like a net positive, especially since it’s open to anyone willing to subscribe. Moreover, we don’t think its design meaningfully changes the fairness of the network.

It’s also great to see the DAO adding another potential revenue stream. With some of the recent changes to infrastructure-level revenues, finding new ways to capture value for the DAO is always welcome, especially when the tradeoffs are limited.

@offchainlabs Thank you for the detailed and transparent response - it’s much appreciated.Overall, your answers address the key concerns effectively. A few constructive points of feedback:

Strengths:

The decision to mirror EIP-4844’s demand-responsive blob fee market for connection pricing is smart. It directly tackles DoS risks with an elegant, market-driven mechanism.

Good explanation on why monopolization becomes economically irrational.

Aligning the fee split (973) with existing Timeboost precedent is a strong governance choice. It reduces friction and shows consistency in how Arbitrum handles ordering-related revenue.

Clarifying that Fast Feed is a pure read layer with no influence on transaction ordering is crucial. This distinction from priority auctions is important for the community to understand.

Areas for further clarity/suggestions:

Transparency & Dashboards - You mentioned public dashboards (Dune, Blockworks) and DAO reporting. Would it be possible to commit to a specific timeline for the first Fast Feed revenue dashboard post-launch? Even a simple Dune dashboard link in the AIP would help build confidence.

Review Cadence — You noted the model will be reviewed periodically. Adding something more concrete (e.g., “reviewed by the DAO every 6 months or upon reaching X connections/revenue threshold”) would strengthen the response.

This proposal feels thoughtful and well-aligned with Arbitrum’s values of permissionless access and credible neutrality. With the above minor enhancements.

Best,

Arb_junior

The following reflects the views of L2BEAT’s governance team, composed of @krst and @Manugotsuka, and is based on their combined research, fact-checking, and discussion.

We voted FOR.

We support the introduction of the Fast Feed. It creates a paid and open access layer for latency-sensitive participants while routing the resulting revenue back to the DAO Treasury and the Arbitrum Developer Guild.

The key point for us is that the feed does not change transaction ordering, inclusion, or the fees paid by regular users. It gives subscribers earlier visibility into transactions after they have already been ordered by the sequencer. That can create some MEV opportunities, but it does not appear to enable frontrunning or sandwiching, which would be the more concerning outcome.

Before voting, we asked our Research Team to review the proposal. Their view was that the design looks good overall, with the same nuance: the Fast Feed may create some MEV opportunities, but not the harmful frontrunning-style behavior that projects generally try to avoid.