Many Eyes - Citizen Science Platform with Expert Verification and Research-Ready Data

Dashboard, projects, project overview, verification queue and verification workspace. Each of the 9 projects and all 23 demo observations, species records and kit readings alike, opens its own page.

The workspace shows what the rule needs and what it has now: 2 agreeing identifiers, not counting the observer, including 1 expert, 67% agreement, to species. Machine suggestions are labelled and never count.

Observations turn amber after 14 days in the queue and red after 30, on the queue cards and in the dashboard's backlog per taxon group, so the oldest records get a verifier first.

5 project types from butterfly counts to river water kits, 23 observations, 18 taxa, 12 volunteers (3 under 18), 9 staff users, 5 upcoming events and 7 needs-attention items.

Verification Without a Record

Experts checked identifications in photo folders, with no record of who agreed with whom and no rule for when a record counts. Without a consensus rule per project, researchers could not tell a confident record from a guess.

Locations That Can Do Harm

Exact points for threatened or heavily collected species, such as some orchids and breeding raptors, can put them at risk. SANBI publishes guidance and a sensitive species list, and NEM:BA regulates threatened and protected species, so public and exported points have to be generalised.

Children and Personal Information

Under POPIA a child is anyone under 18, and processing their information generally needs a competent person's prior consent. Adults who lead school groups or youth events should be checked against the Child Protection Register and the National Register for Sex Offenders first.

Many Protocols, One Dataset

A 15-minute butterfly count on a 1 km route, a tree health survey, sky brightness in mag/arcsec² and river pH from a loaned kit each need their own fields and range checks. Inconsistent protocols, dates and locations had made most of the data unusable for the university.