Summary

  • One-time visitors performed 37% of the work reported in an early 2001 Clickworkers paper; other volunteers kept contributing for weeks.
  • Four clicks around a crater produced a measurement that could be combined with independent observations. Training, comparison and computing capacity made those short visits useful.

They came, clicked on Mars and never returned. In the early Clickworkers experiment, these apparently disappointing visitors did 37% of the work. A website judged entirely by its returning users would have needed a stern meeting. The researchers had crater measurements to inspect.

The figure appears in a 2001 Lunar and Planetary Science Conference paper by B. Kanefsky, N. G. Barlow and V. C. Gulick. Their pilot invited people to identify craters in planetary images through a web browser. Some enthusiasts worked for weeks. The one-visit crowd also left something the team could use. The paper reports the early experiment, with its sequence of quantities running from December 2000 into January 2001.

The task had a small, visible finish line. Click four points around a crater’s rim. The browser fitted a circle, smoothed out some imprecision and rejected inputs that were too far from circular. Submitting the result sent the crater’s latitude, longitude and diameter. A visitor could complete one useful unit without signing up for a second career in planetary science.

That brevity required preparation. A training image contained seven known craters, with feedback, hints and demonstrations to help the newcomer learn the job. Another task asked people to classify craters as fresh, degraded or ghost, supported by definitions, examples and an optional erosion animation. The project spent effort making the work understandable before inviting strangers to do it.

It also gave the same ground more than one observer. Independent markings could be grouped and checked against each other. The early comparison used Viking maps covered by Barlow’s existing catalogue of 42,284 craters. The team therefore had a way to examine what the crowd was finding and missing. Faint craters proved harder to spot, a reason to care about repeated inspection rather than admire the raw number of clicks.

A NASA Ames account published on 26 March 2001 described how five-minute contributions could add up. It likewise reported that more than a third of the work came from people who visited once and never came back. Participants could work with Viking mosaics or higher-resolution Mars Global Surveyor images. Their marks joined a shared calculation after their browser session ended.

This arrangement made leaving quite compatible with contributing. The scientific team chose the task, supplied the examples and organized the checks. The visitor decided how much attention to give. A finished observation could move into the common pool without demanding a continuing relationship with its author. The official NASA record of the conference paper preserves that early experiment in public participation in crater surveying and classification.

There was a less romantic member of the team. The March account said the experiment ran on one old web-server machine pulled from a shelf. Its project leader expected computing resources, rather than a shortage of data, to become a constraint. Gathering brief contributions still left someone with the job of processing them.

Clickworkers had found useful work in a short visit. The four clicks could survive the departure; the checks could give them value. People were free to go. The old server had rather less choice.