How scoring works
A member’s reputation is a single number from 0 to 1000, built from four weighted components and reduced by any penalties. The model is transparent and identical everywhere — the bot and the website compute it from the same code.
The formula
Base (0–100) = Quality·0.35 + Validation·0.30 + Impact·0.20 + Consistency·0.15
Score (0–1000) = clamp( Base × 10 − penalties , 0 , 1000 )Each of the four components is itself a 0–100 value. They’re combined by the weights above into a base score, scaled to the 0–1000 range, then any penalties are subtracted. The result is clamped so it never goes below 0 or above 1000.
1. Quality — 35%
Scored from the message itself, with no input from anyone else. dcRep looks at signals of substance: length past a sensible threshold, code blocks, links, lists and step-by-step structure, and whether the post fits the channel it’s in (a detailed answer in a help channel counts for more than the same text in off-topic). Low-effort and near-duplicate posts are damped.
2. Validation — 30%
Recognition from the community. Reactions and endorsements both count, but they’re weighted by who gives them — a reaction from a high-reputation member carries more than one from a brand-new account. Moderators can give a special “high value” endorsement. To keep this honest, endorsements are capped per member per day and suspicious patterns are discounted (see Fairness & anti-gaming).
3. Impact — 20%
Did the contribution actually move things forward? Impact rewards posts that spark discussion, get meaningful replies, reach a wide thread, or resolve a question. A help-channel answer that ends up marking a thread solved is a strong impact signal.
4. Consistency — 15%
Reputation should reflect a habit, not a single big day. Consistency tracks quality participation across a rolling 30-day window, rewarding members who show up regularly. It’s aware of vacation mode, so planned breaks don’t punish you.
Penalties
Confirmed gaming or abuse subtracts points from the final score. Some penalties are automatic (e.g. spam bursts or duplicate flooding) and some are applied by moderators. Many are reversible on appeal — members can contest a penalty with /appeal, and a moderator reviews it.
How fresh is a score?
Scores recompute continuously. Posting, reacting, endorsing, and thread activity feed a queue that recalculates the affected member’s score; daily and longer-running jobs handle decay, archetype assignment, and leaderboards. In practice a member’s number reflects their recent behavior within moments of it happening.