EIP-15: Social Impact Boost v2

Summary

This proposal extends the existing social impact boost in two directions. First, it introduces a weighted model where different nodes ("people") on the social graph contribute differently based on their own influence, and expands the set of nodes the graph considers beyond what it uses today. Second, it opens the boost to profiled users at 50% of the non-profile ceiling — so users who have already invested in Ethos can leverage the social signal they built off-platform, instead of losing access to it entirely when they claim.

Motivation

The current social impact boost (max +400) is only available to unclaimed accounts. When a user claims an Ethos profile, the pre-claim boost enters a 60-day linear decay and permanently zeroes out — the user loses access to any future social-graph credit entirely. That creates two problems:

• It punishes the users investing their time in Ethos. A profiled member with substantial off-Ethos social signal gets nothing from it after decay runs out, while an unclaimed account with equivalent signal keeps up to +400 indefinitely. The incentive is inverted.
• It treats every node identically. Signal from high-credibility accounts who don't participate in Ethos is almost certainly cleaner than signal from an Ethos-native node whose classification depends on the same system it's feeding into. The v1 system can't distinguish between them, and influence factor has been bloated by circular nodes of power within Ethos itself.

We still believe Ethos's core thesis requires explicit signals — reviews, vouches, slashes — to dominate the score. But a well-measured external social graph is valuable as a small percentage of overall credibility for both profiles and non-profiles, and today we're capturing almost none of it for the users most invested in the network.

Specification

• Weighted graph contribution. Different nodes contribute at different rates based on node type and classification. Algorithmic details are intentionally not published, but the direction is that nodes carrying independent, off-Ethos signal count meaningfully more than nodes whose signal is shaped by internal Ethos incentives. The final contribution is percentile-ranked across all targets and mapped into the existing 10–400 boost range.
• Expanded graph coverage. The set of nodes the calculation considers is broadened beyond the sources used by v1, so users with real off-Ethos credibility can reach higher tiers without first building an Ethos-native presence.
• Profile eligibility at 50% cap. Profiled users become eligible for the live weighted boost, not just the 60-day decay of a pre-claim snapshot. The ceiling for profiled users is 50% of the non-profile ceiling — up to +200 at top-tier influence, roughly +100 at mid-tier. Non-profile accounts continue to use the full 0–400 range, retaining a structural advantage to compensate for their lack of access to vouches and mutual-vouch signals.
• Influence misalignment penalty still applies. The social boost contributes to your score, but users whose social reach is disproportionately low relative to their verified Ethos activity (reviews, vouches, etc.) will continue to see an offsetting penalty. Social signal is amplified at the top end, but it's always weighed against whether you've earned comparable credibility through explicit Ethos mechanisms.

Risk

The largest new attack surface is social graph hijacking — a bad actor takes over an aged account with a rich graph history, claims an Ethos profile, and inherits up to +200 of undeserved credibility. This is real but not novel: the same risk already exists in claimed-profile decay today (pre-claim signal is captured at claim time and persists for 60 days), and it applies symmetrically to any identity-reuse attack on Ethos.

Mitigations:

• Slashing. A reviewer who identifies a hijacked profile can initiate a slash that strips the score.
• Score cap. +200 is material but doesn't single-handedly move a user into a higher tier, and other score elements still apply a corrective when social signal is inflated relative to verified Ethos activity.

Impact

• Current Ethos users gain. Top-tier Ethos profiles with genuine off-Ethos signal unlock up to +200 of previously-zeroed score. Mid-tier gains ~+100. Users with no external signal are unaffected.
• Non-profile accounts shift under the new weighting. Some accounts whose signal is dominated by Ethos-internal activity will see their boost decrease; accounts with strong off-Ethos signal will see their boost increase. Scores for well-known external accounts remain at or near the ceiling.
• Non-profile advantage preserved. Non-profiles retain a 2x higher ceiling than profiles (400 vs 200), reflecting the fact that they can't access vouches or mutual-vouch signals.
• Core thesis preserved. Social impact remains a minority contributor to the overall score; reviews, vouches, and slashes continue to dominate.

Future Considerations

• Raise the profile cap above 50% if v2 rollout shows the signal is net positive.
• Integrate additional graph sources once they reach sufficient density to provide meaningful signal.
• Incorporate additional measures of influence like Github network data.