Plain Language Summary: We are running a scientific experiment to find out how random real card shuffles are. We are asking people to video themselves shuffling a standard 52-card deck and then revealing the cards one by one. Our AI reads the card order from your video, you check it's correct, and we add the card order (as a list of numbers) to a research database. We compare every shuffle against every other shuffle to see how close any two get. We strip out all personal information and keep the card-order data forever for research. This page explains everything you need to know before taking part.
Part 1: Participant Information Sheet
1. Invitation to Participate
You are being invited to take part in a research study. Before you decide whether to participate, it is important that you understand why the research is being done and what it will involve. Please take time to read the following information carefully. You are welcome to discuss it with others and to ask us questions if anything is unclear or if you would like more information.
This study investigates a simple but scientifically interesting question: how random are real human card shuffles?
There are 52! (52 factorial) ways to arrange a standard deck of playing cards. That number is approximately 80,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 — far more than the number of atoms in the observable universe. Mathematicians have predicted that casual shuffling produces far less randomness than people assume, but this has never been tested at scale with real people.
We are building the world's first large-scale, video-verified dataset of real human shuffle permutations. By collecting thousands of shuffles, we aim to:
Measure how random real shuffles actually are, compared to the mathematical ideal;
See how close any two independently shuffled decks have ever come to matching;
Study whether different shuffling techniques (riffle, overhand, wash) produce different levels of randomness;
Create a public dataset for other researchers in mathematics, statistics, cognitive science, and computer vision.
You have been invited because you have accessed the 52Shuffled website and expressed interest in participating. We are seeking a large and diverse group of participants — anyone aged 16 or over who has access to a standard 52-card deck and a device with a camera can take part. There are no other selection criteria. We aim to collect submissions from as many people as possible, across different ages, locations, and shuffling styles.
No. Participation is entirely voluntary. You are free to decide whether or not to take part. If you do decide to take part, you will be asked to confirm your consent (see Part 2). You can withdraw at any time without giving a reason and without any negative consequences. Not taking part will not affect your ability to use the non-research features of the 52Shuffled website.
If you decide to take part, you will be asked to do the following:
Step 1 — Set Up
You will need a standard 52-card deck (no jokers) and a device with a camera (smartphone, tablet, or laptop). The 52Shuffled web app will open your camera.
Step 2 — Record Your Shuffle (Video)
The app will guide you through a three-phase video recording:
Fan Phase (about 5 seconds): Show all 52 cards fanned out face-up to the camera. This proves you have a complete, standard deck.
Shuffle Phase (20–30 seconds): Shuffle the deck in whatever way you like — riffle, overhand, wash, or any combination. The app will tell you when to stop. The shuffle duration is randomly varied (you do not choose it) to prevent pre-arrangement.
Reveal Phase (about 60 seconds): Go through the deck card by card, showing each card face-up to the camera, from the top of the deck to the bottom.
Step 3 — AI Card Recognition
Our AI system will analyse your video and attempt to identify each of the 52 cards in the order you revealed them. This takes a few seconds.
Step 4 — Verification
You will be shown the AI's reading of your deck and asked to check it. If any cards were misidentified, you can correct them using a simple tap-to-fix interface. This step is important — your corrections ensure the data is accurate and also help us improve the AI.
Step 5 — Results
Once confirmed, your card order is compared against every other shuffle in the database. You will see:
Your closest match (how many cards landed in the same position as another participant's shuffle);
Your percentile (how your closest match compares to others);
The global record (the closest pair of shuffles ever found);
A shareable result card.
You can submit as many shuffles as you like (subject to daily rate limits for data quality reasons). Each submission follows the same process. You do not need to use the same deck each time.
This is a low-risk study. The potential risks and disadvantages are:
7.1 Privacy
Important — Video Content: Your video recording must show only your hands and the playing cards. Do not include any other persons, faces, identifiable features, or personal items in the video. Do not film in locations where sensitive or identifiable information is visible in the background. This protects your privacy, as videos are retained for the full duration of the experiment.
Your video recording will show your hands shuffling cards. Videos are retained for the full duration of the experiment (which may be several years) as verification evidence. If two shuffles produce a close or exact match, video review is essential to confirm authenticity and rule out cheating. The video is not published, shared publicly, or included in the research dataset. Only the numerical card order (a list of 52 numbers) is kept for research — this cannot identify you.
7.2 Time
Each submission takes 2–4 minutes. This is a minor time commitment. You are free to stop at any time.
7.3 Data Permanence
Once your card order data has been anonymised and added to the research dataset, it cannot be individually removed (see Sections 10, 11, and 15 for full details). You should understand this before participating. However, since the retained data is just a list of 52 numbers with no link to your identity, it poses no privacy risk.
7.4 Minor Frustration
The AI card recognition may occasionally misread cards. You may need to correct a few cards during the verification step. This is usually straightforward.
There are no known physical, psychological, financial, social, or legal risks associated with participating in this study beyond those described above.
You will see how your shuffle compares to thousands of others;
You will learn about the mathematics of card shuffling and permutations;
You will receive a personalised result showing your shuffle's statistical properties;
You contribute to a genuine scientific dataset — your participation has real research value.
8.2 Benefits to Science and Society
This dataset has never existed before — it will advance our understanding of human randomness behaviour;
It will help validate or challenge long-standing mathematical predictions about shuffle quality (notably the Diaconis–Bayer theorem that 7 riffle shuffles are needed for randomness);
The data may benefit AI and computer vision research (card recognition);
The dataset will be made available to other researchers under appropriate licensing.
8.3 No Payment
There is no payment, prize, or financial compensation for taking part. Your participation is voluntary and in the interest of contributing to research.
We collect two types of data. They are handled very differently:
9.1 Personal Data (linked to your identity)
Data
What happens to it
Name and email
Used for your account. Deleted when you delete your account (+ 12 months for legal compliance).
Video recording
Used for AI card recognition, anti-fraud checks, and match verification. Retained for the full duration of the experiment (which may be several years). Videos may be anonymised (faces and identifiable features removed or obscured, if any) and supplied to third-party researchers for additional research purposes (e.g., computer vision, card recognition AI training). Unanonymised videos are never published or shared publicly.
Snapshot images
Displayed on your results page. Deleted with your account.
IP address
Used for rate limiting and security. Retained for the duration of the research + up to 5 years after completion.
Device information
Used for anti-fraud. Retained for the duration of the research + up to 5 years after completion.
9.2 Research Data (anonymised — no link to you)
Data
What happens to it
Card order (52 numbers)
Permanently included in the research dataset. Cannot identify you.
Statistical metrics
Displacement score, fixed points, Kendall tau distance, etc. Derived from the card order. Permanently retained.
Comparison results
How your shuffle compared to others (number of matching positions). Permanently retained.
Quality tier
A confidence rating for data integrity (Gold/Silver/Bronze). Permanently retained.
Submission timestamp
May be generalised (e.g., to month/year only) before inclusion in the dataset.
The key point: The research dataset contains only numbers and statistics. It does not contain your name, email, video, images, IP address, or any information that could identify you.
Before any data enters the research dataset, we apply the following anonymisation steps:
Remove your name, email, and profile image — these are permanently stripped.
Replace your user ID — your internal database ID is replaced with a random, non-reversible code. We destroy the mapping between the random code and your real ID.
Delete your IP address and device fingerprint — these are not included in the dataset.
Exclude your video and images — only the numerical card order and derived statistics are kept.
Generalise timestamps — precise times may be rounded to prevent re-identification.
Check for re-identification risk — we assess whether any combination of remaining data points could identify you. If so, we apply further measures.
After anonymisation, the data cannot be linked back to you by us or by anyone else. This is irreversible by design.
Full duration of the experiment (which may be several years) — required for match verification
IP address
Duration of the research + up to 5 years after completion
Device fingerprint
Duration of the research + up to 5 years after completion
Anonymised Research Data
Data
Retention
Card order (52 numbers)
Indefinitely (forever)
Statistical metrics and comparisons
Indefinitely (forever)
Important: Anonymised data is kept forever as part of the scientific record. This is necessary for the integrity and reproducibility of the research. Since this data cannot identify you, its permanent retention does not pose a privacy risk. However, please be sure you are comfortable with this before participating.
Your personal data (name, email, etc.) will not be shared with other researchers, commercial entities, or the public. It is shared only with our essential service providers (cloud hosting, AI card recognition) under strict data processing agreements, and only to the extent necessary to operate the service.
Video recordings may be anonymised (faces and identifiable features removed or obscured, if any) and supplied to third-party researchers for additional research purposes, such as computer vision research, card recognition AI training, and shuffle technique analysis. Anonymised videos are stripped of all metadata linking them to your identity.
Anonymised Research Data
The anonymised dataset (card orders and statistics only — no personal information) may be shared with:
Academic researchers — for studies in mathematics, statistics, cognitive science, and computer vision;
Educational institutions — as a teaching resource for probability and combinatorics;
Other qualified organisations — under licensing agreements that prohibit any attempt to re-identify participants.
All recipients of the dataset are contractually required not to attempt to identify any individual from the data.
Yes — but only the anonymised dataset, never your personal data.
We may generate revenue from the anonymised research dataset in the following ways:
Licensing access to the dataset for commercial research (e.g., gaming companies benchmarking their shuffle algorithms);
Providing paid API access for developers and researchers;
Licensing AI training data (card recognition models) derived from aggregated, anonymised user corrections;
Publishing research papers and reports.
You will not receive any payment or share of revenue from the commercialisation of the dataset, regardless of how much data you contribute. Your participation is voluntary and in the interest of scientific research. Revenue generated supports the ongoing operation of the project and further research.
We are transparent about this because we believe you should know exactly how your contribution will be used before you decide to participate. If you are not comfortable with the potential commercial use of anonymised data, please do not participate.
A self-hosted computer vision system (YOLOv8) analyses your video to identify each card. Your video is processed entirely on our own servers — it is not sent to any third-party AI service. The AI returns a list of card identities. Video data is retained only as described in Section 11.
14.2 Improving Accuracy
When you correct AI mistakes during the verification step, those corrections help us understand where the AI struggles (e.g., certain lighting conditions, card designs, or angles). Over time, this improves recognition accuracy for everyone. Your corrections are stored in anonymised form and may be used as training data for card recognition models.
14.3 Automated Decisions
We use automated systems to detect potentially fraudulent submissions (e.g., pre-arranged card orders, statistically impossible results). If your submission is flagged by the automated system, it will be reviewed by a human before any action is taken against your account. You will not be penalised solely on the basis of an automated decision.
AI is a tool, not a decision-maker. The AI reads cards from your video, but you verify and confirm the final card order. The authoritative data is what you confirm, not what the AI initially detects.
You can delete your account at any time by contacting us;
Upon account deletion, we will delete your personal data (name, email, profile, video recordings, images, IP address, device information);
You can stop participating at any time without giving a reason;
Withdrawal will not result in any penalty or disadvantage.
What Cannot Be Withdrawn
Anonymised card order data that has already been incorporated into the research dataset cannot be individually identified or removed. This is because:
The link between your identity and the data has been permanently destroyed;
We cannot determine which records in the anonymised dataset were yours;
Removing individual records would compromise the integrity of the research;
This is consistent with the UK GDPR research exemption (Article 17(3)(d)) and the Data Protection Act 2018 (Schedule 2, Part 6, Paragraph 27).
Please decide carefully before submitting. Once your card order data has been anonymised (which happens shortly after you confirm your submission), it becomes part of the permanent research record. We want you to understand this fully before you participate. If this concerns you, please do not submit a shuffle.
How to Withdraw
To withdraw from the study and/or delete your account, contact us at contact@52shuffled.com with the subject line "Withdrawal Request." We will process your request within 30 days.
Under the UK General Data Protection Regulation (UK GDPR), you have the following rights in relation to your personal data:
Right of access — you can request a copy of the personal data we hold about you;
Right to rectification — you can ask us to correct inaccurate data;
Right to erasure — you can ask us to delete your personal data (subject to the limitations described in Section 15);
Right to restrict processing — you can ask us to limit how we use your data;
Right to data portability — you can request your data in a machine-readable format;
Right to object — you can object to processing based on legitimate interests;
Rights related to automated decisions — you have the right not to be subject to decisions based solely on automated processing that significantly affect you.
To exercise any of these rights, contact us at contact@52shuffled.com.
These rights apply to your personal data only. Anonymised data (which cannot identify you) is not covered by these rights.
This is an independent research project with a hybrid sustainability model:
Advertising Revenue: Non-intrusive display advertisements are shown during processing/waiting screens only (never during the experiment phases). This provides the primary funding for server infrastructure and storage costs. Participants may opt out of advertisements by subscribing to Premium membership.
Premium Membership: An optional paid subscription offering enhanced features and an ad-free experience. Premium membership does not affect research participation or data quality—all participants contribute equally to the research dataset.
Data Licensing: In the future, commercial entities may license access to the anonymised dataset for product development or AI training. Academic and non-profit research access remains free or controlled via Data Use Agreement.
Independence Statement: Advertising partners, premium members, and future commercial licensees have no role in study design, data collection, analysis, or publication decisions. All research methodology decisions are made independently by the research team. No advertiser or sponsor has access to non-public participant data or editorial control over research findings.
Conflicts of Interest: The research team has no financial relationships with card manufacturers, casino operators, or gaming companies that could bias study design or results.
This study has been subject to an internal ethical self-assessment. We have determined that the study poses minimal risk to participants (no sensitive personal data is retained in the research dataset, no vulnerable populations are specifically targeted, participation is voluntary, and full informed consent is obtained). We have designed the study in accordance with the principles of the Declaration of Helsinki, the ESRC Framework for Research Ethics, and the UK GDPR.
A full research protocol document is available upon request.
Study Title: Empirical Characterisation of Human Card Shuffle Randomness: A Large-Scale Crowdsourced Permutation Dataset
Lead Researcher: S Singh
Organisation: 52Shuffled Ltd
Participant ID: [Auto-assigned / entered at sign-up]
Please read each statement below carefully. For the digital version, tick each box to confirm your understanding and agreement. Items marked with * are required to participate. Items marked (optional) are your choice — you can participate without agreeing to these.
Required Consent
Optional Consent
Declaration
For the digital version of this form, ticking the boxes above and clicking "I Consent" constitutes your electronic signature and has the same legal effect as a handwritten signature. For the paper version, please sign below.
For participants aged 16–17:
Researcher use only:
* Required — you must agree to all required items to participate.
(optional) — your choice. You can participate without agreeing to these items.
One copy of this form should be retained by the participant. One copy should be retained by the research team.