How Picknado works

Start with the task

Picknado has six tools for everyday choices. Use a wheel to select one option, a coin for two equally likely alternatives, dice for game rolls, a number generator for numerical samples, a list shuffle for a complete order, and Team Generator for groups or changing partners. Names and Yes / No are modes of those tools, not separate algorithms.

Agree on acceptable options before drawing. Random assignment does not account for accessibility, ability, urgency or personal preferences. For a workshop, review the group’s needs first and use the schedule as an organising aid.

Where the random values come from

The tools request unsigned 32-bit values from your browser’s Web Crypto getRandomValues API. They do not use Math.random, the time of your click, or an AI model to choose a winner. MDN describes this browser random source and its cryptographic properties.

Why we discard some random values

Mapping a source to a smaller range can introduce bias. Imagine ten equally likely source values, 0–9, and three choices. Taking the remainder after division by three assigns four values to choice 0, but only three to choices 1 and 2. Discarding 9 and trying again leaves three source values per choice. Picknado applies this rejection method to its actual 32-bit source before choosing an integer.

Lists, dice and animations

Lists use Fisher–Yates: work through the list, choose a position uniformly from the remaining positions, and swap. Every entry remains present exactly once. Identical lines remain separate entries, and an unchanged order is a possible result. The NIST description of Fisher–Yates explains the method.

The wheel selects an entry before animating to it. Skipping the animation changes the waiting time, not the selection method. Each coin flip and each die uses a fresh selection. Past outcomes do not make a future outcome due. Unique numbers are sampled without replacement within one draw; a later draw starts afresh.

How partner rotation prevents repeat pairs

Ordinary teams start with a shuffled list and distribute entries across balanced groups. Rotation uses a different method: shuffle the participants once, arrange their positions in a ring, hold one position fixed and rotate the others between rounds. Opposite positions form pairs. This creates a complete cycle in which each unordered pair appears once.

With an odd group, an extra empty position represents a rest turn. A full cycle has n−1 rounds for even n, or n rounds for odd n. Each person in a complete odd cycle rests once. A shortened cycle has no repeated pair, but rest turns need not be equal.

The random start varies the schedule; this is not a promise to sample every mathematically possible schedule equally. A new plan does not remember previous meetings. Fixed partners, exclusions and balancing by experience are not supported. See a five-person example and a practical session plan.

What stays in the browser

Our tools process input lists and results in the open tab. They do not upload that data, put names into links, or automatically save results. Reloading clears the session. Copy, Share and Download CSV act only when you request them; exported copies remain outside the page until you remove them.

This does not mean that visiting the website creates no network traffic. Hosting and security providers receive technical requests. The homepage and tool pages load Google advertising and consent scripts. Optional animation preferences are stored locally. The privacy notice describes these separate processes and controls.

What we check, and what we do not certify

Automated tests check integer boundaries, uniqueness, list preservation and rotation schedules for groups of 2–40 people, including odd-group rest turns. Browser tests exercise the actual controls, errors, keyboard use, mobile layouts and local exports. Worked examples explain conditions that you can check in a result yourself.

These checks are not independent certification of randomness. Picknado has no public draw audit log, identity verification or saved attendance record. It is intended for everyday decisions and group activities, not regulated prize draws or security credentials.

If a result violates a stated rule, use the contact details in the legal notice and describe the tool, settings and browser. Replace real participant names with fictional examples. There is no need to email the original list.