The floor and the usual range
One large regional fair's category judging guide, listed below, requires that every project eligible for a category award be evaluated by at least two judges, and organises judges into teams of two to four that each see eight to twelve projects at about eight to ten minutes per project. A hackathon organiser guide from the same list recommends that every project be seen by at least three different judges. Fair networks also ask fairs to tell students roughly how many interviews to expect, which is exactly the judges per project number.
The capacity check
Every project in every round needs its panel. With P projects, J judges per project and R rounds, the day holds P times J times R judge visits. Each judge can take up to their maximum. Add the maximums: if the total is smaller than the visits, the fair is short before any conflict is counted. Conflicts, categories and time windows only make it tighter, so leave a margin of ten to twenty percent.
Worked example: two, three and four judges per project
The sample fair in the judge assigner has 36 projects and 12 judges with a maximum of 10 each, so the judges can take 120 visits between them. The table shows the assigner's result for each panel size with everything else unchanged.
| Judges per project | Seats | Filled | Projects short | Judge loads |
|---|---|---|---|---|
| 2 | 72 | 72 | 0 | 6 each |
| 3 | 108 | 108 | 0 | 8 to 10 |
| 4 | 144 | 120 | 20 | 10 each |
With four judges per project the fair needs 144 visits and the judges can give 120, so 24 seats stay empty across 20 projects. The assigner fills 120 and says that no arrangement can fill more under these rules, because its maximum flow check proves the ceiling. The fix is not a cleverer schedule: it is more judges, higher maximums or smaller panels.
How many projects per judge
Work back from the time you have. A two hour judging block at eight to ten minutes per interview, plus a minute or two of walking, gives room for about eight to twelve visits; build in a break and six to ten is kinder. Set the fair maximum to that, and give individual judges a lower maximum if they are only there for part of the session. The minimum per judge is a target, not a rule: the assigner tries to give every judge at least that many, and says who fell short.
Common mistakes
- Recruiting judges by headcount instead of by visits needed.
- Giving panels in one category four judges and panels in another two, then comparing their scores.
- Setting one fair wide maximum for judges who arrive at different times.
- Forgetting the second round when counting visits.
- Discovering a short panel at noon instead of from the plan the week before.
For keeping judges in their own categories while you fill the panels, see category and expertise matching.