Hi Edil, Appreciate taking the time to review the BIP. > But I'm not so sure about how that improves the main purpose of the network. Quite the contrary, seems to open more DoS and fingerprinting vectors. For DoS, the concern is whether a small message can cause disproportionate processing, disk usage, or outgoing traffic. Merely sending invalid data is not a new DoS vector; peers already have many ways to do that. It's not possible to prevent all of them. Currently, I don’t see any resource-exhaustion or bandwidth amplification vector introduced by this proposal. Processing is bounded by the branch-length, recency and retained-tip limits. Nodes are not required to download stale block data. Nodes that choose to collect it use the normal block-download mechanisms and resource limits. Implementations may also apply stricter limits, rate-limit abusive peers, or disable the feature. Also on mainnet, producing stale headers requires real proof of work comparable to the active chain. On fingerprinting - supporting staletip does add another observable characteristic. This is a trade-off the node makes when enabling the feature and is common to optional P2P features generally. Nodes for which this trade-off is unacceptable can disable staletip relay. > About the design, it's not clear to me how that could work in face of more than one competing chain. Each staletip message describes one linear branch. Assuming A1-A2-A3-A4 is the active chain, the two competing branches would be announced separately: staletip(fork_point=A2, headers=[B3, B4]) staletip(fork_point=A3, headers=[C4]) The protocol announces competing branches in separate messages. I’ve clarified this in the BIP draft. Thanks for pointing it out :) Thanks again for reviewing. Hope this addresses your concerns. Feel free to reach out or leave additional comments. Cheers, Ram On Tuesday, August 11, 2026 at 7:04:41 AM UTC-7 Edil Guimarães de Medeiros wrote: > There's an ongoing effort to document mainnet stale blocks: > https://github.com/bitcoin-data/stale-blocks > As one can imagine, since there is currently no way to recover them from > the network itself, this > requires constantly monitoring several long running nodes in the hope to > get them. From a monitoring > and research perspective, this proposal could be useful. > > But I'm not so sure about how that improves the main purpose of the > network. Quite the contrary, seems > to open more DoS and fingerprinting vectors. > > About the design, it's not clear to me how that could work in face of more > than one competing chain. For > instance, suppose I have: > > /- C4 > A1 - A2 - A3 - A4 > \- B3 - B4 > > It's not clear to me how the proposal deals with communicating this, which > is quite common in signet and > the testnets, since each message is designed to relay a single competing > branch (e.g. B3-B4 above). > > Left more detailed and editorial comments in the bip repository PR. > > Cheers. > Edil > > Em qua., 29 de jul. de 2026 às 14:28, Ram escreveu: > >> Hello list, >> >> This proposal introduces `staletip`, an opt-in P2P message for relaying >> recent >> stale tips between peers. Building on AJ Towns' initial work, w0xlt and I >> have >> developed the proposal further and built a proof-of-concept >> implementation. >> >> Draft BIP: >> >> https://github.com/pseudoramdom/bips/blob/staletip-bip-draft/bip-staletip.md >> >> Proof-of-concept: >> https://github.com/w0xlt/bitcoin/tree/staletip-v4 >> >> Today, once a block loses a race and goes stale, it stops propagating – >> compact >> block relay and FIBRE aggressively relay the winning chain, while stale >> branches fall away. That's great for fast propagation, but it makes the >> stale >> rate difficult to observe. Even dedicated monitors [0] have only partial >> views: >> a stale block seen by one monitor may never reach another. >> >> The stale rate is a useful network-health signal because it is closely >> related >> to block propagation delay. The longer it takes miners to learn about a >> newly >> found block, the greater the chance that another valid block will be >> found at >> the same height, creating a race in which one of the blocks becomes stale >> [1]. >> An elevated stale rate could also expose validation or relay bottlenecks. >> For example, the May 2023 "inv-to-send" bug degraded block propagation and >> coincided with a roughly 10x increase in the observed stale rate [2]. >> >> A stale block does not by itself identify its cause, but changes in the >> rate or shape of stale branches can provide a reason to investigate: >> >> - a network partition – when a partition heals, blocks mined on the >> losing side become stale, potentially producing several related stale >> blocks at once. >> - adversarial mining strategies such as selfish mining [3] – these >> can cause honest miners' blocks to become stale. >> >> The proposal fills this observability gap with an opt-in P2P message >> called >> `staletip`, allowing nodes to proactively announce recent stale tips to >> peers. >> >> An added benefit is potentially faster reorg handling: if a node already >> knows >> the relevant headers, and perhaps has the block data, it has less work to >> do if >> that branch later becomes active. >> >> At protocol level, nodes advertise support using BIP 434 `feature` >> message [4], >> and `staletip` messages are only sent to peers that advertised support. >> Each >> announcement contains the fork point, a sequence of compressed headers, >> and a >> flag indicating whether the sender can serve the stale tip block. >> >> The proposed default relay policy: >> - relays only tips within 1000 blocks (about seven days) of the active >> tip, >> keeping announcements recent and stale-tip spam costly on mainnet. >> - limits branches to 20 compressed headers, covering short-term reorgs >> without tracking persistent chain splits and keeps each announcement >> under >> 1 kB. >> - keeps at most 10 recent tips in the relay cache, so announcing the >> full >> cache to a new peer remains under 10 kB, excluding any blocks >> requested. >> >> The BIP text has more background and the full message format. Comments are >> welcome. >> >> Cheers, >> Ram (pseudoramdom ) >> & >> w0xlt >> >> [0] https://github.com/bitcoin-data/stale-blocks >> [1] >> https://delvingbitcoin.org/t/propagation-delay-and-mining-centralization-modeling-stale-rates/2110 >> [2] https://b10c.me/observations/15-inv-to-send-queue/ >> [3] https://arxiv.org/abs/1311.0243 >> [4] https://github.com/bitcoin/bips/blob/master/bip-0434.md >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Bitcoin Development Mailing List" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to bitcoindev+...@googlegroups.com. >> To view this discussion visit >> https://groups.google.com/d/msgid/bitcoindev/d92f1615-368b-4406-b326-a1799c72a555n%40googlegroups.com >> >> . >> > > > -- > Edil > -- You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group. To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/7bbef0df-dbb5-48d1-9671-b0cec91fcbb0n%40googlegroups.com.