The Honest Metrics ledger
Standing here is built on evidence, not attention. This page states plainly what we count, how we count it, and what we refuse to count, so you can hold the product to it.
A number you cannot verify is a number you should not trust. Every figure on Skills Tech Network should trace to a real, attributable signal, or it should not exist.
Last updated July 20, 2026.
How view counts work
Views are a real, deduplicated count of distinct signed-in engineers, not raw impressions.
A view is counted only from a signed-in account, once per person.
Each view is keyed by a salted SHA-256 of the viewer's account, so the same person viewing repeatedly counts once. Logged-out traffic is never counted.
We never store your email against a view, and we never count self-views.
Only the salted hash is stored, never the email. A view of your own post or profile is excluded before it is recorded.
View counts make no external calls and cannot be inflated.
Counts are computed from first-party rows only. There is no bot traffic, no seeding, and no third-party analytics inflating the number.
Only you see your own view counts.
Reach numbers are shown to the owner of a post or profile, not broadcast publicly as a vanity badge.
How reputation works
Proof Reputation is per-skill and earned from verified-useful work, not a single global karma score.
Reputation is contextual: points attach to a specific skill.
A popular joke does not raise your architecture standing. Points come from typed reactions, accepted answers, and peer-reviewed builds, each credited to the relevant skill.
An endorsement is weighted by the endorser's own proven standing in that skill.
A vouch from someone who has themselves proven the skill counts more; an unproven vouch counts little. Endorsements are bounded so they never rival hard proof, and rings cannot farm points.
External proof is shown exactly as it was verified, never upgraded.
Evidence from Skills Tech Talk carries its own honesty category (for example, tool-verified or self-reported), and we render that category as-is. We never promote a weaker claim into a stronger one.
The proof an open question would earn is a published formula, not an engagement number.
The Open Board shows the reputation an accepted answer credits: twelve points per credited skill, capped at three skills. It is the exact amount you would earn, not a figure tuned to make you click.
How moderation works
Ranking is a stated policy, not a hidden hand. Signal is promoted; noise is downranked in the open.
Off-topic, political, spam, and engagement-bait content is downranked, not silently deleted.
The pipeline labels content and lowers its rank. Content that breaks the rules can be removed by an authorized moderator, and every enforcement decision can be appealed.
Fake engagement is a rule violation, not a growth tactic.
No fake users, no fake likes, no inflated views, no pay-to-rank. See the Community Guidelines and Acceptable Use Policy.
What we will not count
Some numbers exist only to manufacture attention. We do not keep them.
No global karma score.
Reputation is per-skill and evidence-backed; there is no single number to farm.
No raw impression counts, no follower vanity totals as a rank signal, no streaks.
We do not reward showing up for its own sake, and we do not turn engagement pressure into a metric.
No purchased or seeded activity, ever.
Every figure traces to a real, attributable action by a real account.
Changelog
A dated record of every change to what we count or how, so this page is auditable over time.
- July 20, 2026First published: deduplicated views, per-skill reputation, proof-weighted endorsements, open downranking, and the do-not-count list.