LumiKin

Methodology

How it works

LumiKin rates games on what they actually do to a developing mind — not just what they contain. Here's exactly how every score is calculated.

faq.sectionLumiScoreBenefit Density Score (BDS)Risk Intensity Score (RIS)Daily time recommendationHow games are reviewedAdditional context fields

faq.sectionLumiScore

What is the LumiScore?+
The LumiScore is a single 0–100 number that summarises how well a game works for a developing child. It is the harmonic mean of two independent scores — the Benefit Density Score (BDS) and a safety score derived from the Risk Intensity Score (RIS). Using the harmonic mean means a game can't hide a very low score on one side behind a very high score on the other.
How is the LumiScore calculated?+
LumiScore = 100 × (2 × BDS × Safety) / (BDS + Safety)

where Safety = 1 − RIS. A game with excellent developmental value (BDS 0.80) but very high manipulation risk (RIS 0.80, Safety 0.20) scores only 32 — reflecting the real tension between what the game teaches and how it keeps kids playing.
What do the score ranges mean?+
  • 70–100 — Recommended. Strong developmental value, low manipulation design.
  • 40–69 — Play with awareness. Worthwhile but has notable risk factors to manage.
  • 0–39 — Use caution. High risk relative to developmental benefit. Keep sessions short.
Why isn't the ESRB rating enough?+
ESRB rates content (violence, language, sexual themes) — not design. A game can be rated E for Everyone and still use slot-machine reward loops, aggressive push notifications, or unlimited in-app purchases targeting children. The LumiScore captures those design decisions, which are invisible to ESRB but highly relevant to parents.

Research

Benefit Density Score (BDS)

What does the BDS measure?+
The BDS (0.00–1.00) reflects how much genuine developmental value a game provides across three categories: cognitive skills, social-emotional skills, and physical/motor skills. A high BDS means the game actively builds skills that transfer outside the game.
What are the three benefit categories?+

B1 · Cognitive development (50% of BDS)

Problem solving, spatial awareness, strategic thinking, critical thinking, memory, creativity, reading, math/systems thinking, real-world learning transfer, and adaptive challenge. Scored across 10 dimensions (0–5 each, max 50 points).

B2 · Social & emotional development (30% of BDS)

Teamwork, communication, empathy, emotional regulation, ethical reasoning, and quality of social interaction. Scored across 6 dimensions (max 30 points).

B3 · Physical & motor development (20% of BDS)

Hand-eye coordination, fine motor skills, reaction time, and physical activity (VR/motion). Scored across 4 dimensions (max 20 points).

Can a casual or simple game score well on BDS?+
Yes, if it genuinely develops skills. Tetris scores well on spatial awareness and reaction time. Stardew Valley scores well on planning and emotional regulation. A game doesn't need to be complex — it needs to build real skills rather than just provide passive stimulation.

Research

Risk Intensity Score (RIS)

What does the RIS measure?+
The RIS (0.00–1.00) captures how aggressively a game uses design patterns that can manipulate behaviour — especially in developing minds. It does not measure content (violence, language) — that is captured separately as Content Risk.
What are the three risk categories?+

R1 · Dopamine manipulation design (45% of RIS)

Variable-ratio reward loops, streak mechanics, loss aversion, FOMO events, artificial stopping barriers (energy systems), re-engagement notifications, near-miss mechanics, infinite scroll design, escalating commitment, and variable reward frequency. Scored 0–3 each across 10 factors (max 30 points).

R2 · Monetization pressure (30% of RIS)

Spending ceiling, pay-to-win mechanics, currency obfuscation (gem → coin → credit), in-game spending prompts, child-targeting design, ad pressure, subscription pressure, and social spending dynamics. Scored 0–3 each across 8 factors (max 24 points).

R3 · Social & emotional risk (25% of RIS)

Social obligations (guild events, daily team commitments), competitive toxicity, stranger interaction risk, social comparison mechanics, identity/self-worth tied to in-game status, and privacy risk. Scored 0–3 each across 6 factors (max 18 points).

What about violence, language, and other content?+
Content risk (R4) is tracked separately and is NOT included in the RIS or LumiScore. It covers violence level, sexual content, language, substance references, and horror intensity — aligned with existing ESRB/PEGI categories. We display it alongside the score as a parental judgment call, not a formula input. A game can be perfectly safe by design (low RIS) but still carry content that is inappropriate for younger children.
Can a game have a high RIS and a high BDS at the same time?+
Yes — and this is an important nuance. Genshin Impact is a good example: it has genuine exploration value, team-based combat, and world-building (moderate BDS), but it also uses gacha mechanics, daily streaks, FOMO banners, and unlimited spending (high RIS). The LumiScore reflects this tension honestly. High-benefit, high-risk games get a special note recommending shorter sessions and a conversation with your child about why the game is designed the way it is.

Research

Daily time recommendation

How is the daily time recommendation calculated?+
The base recommendation comes from the RIS:
RIS rangeBase recommendation
0.00 – 0.15Up to 120 min
0.16 – 0.30Up to 90 min
0.31 – 0.50Up to 60 min
0.51 – 0.70Up to 30 min
0.71 – 1.0015 min or not recommended
Does a high BDS change the time recommendation?+
Yes — but only under specific conditions:
  • If BDS ≥ 0.60 (substantial developmental value), the recommendation extends one tier — unless RIS > 0.70, where high risk overrides the benefit extension.
  • If BDS < 0.20 AND RIS > 0.30 (low value, moderate risk), the recommendation drops one tier.

This asymmetry is intentional: benefits can earn a little more time, but they cannot override a very high-risk design.

Does the recommendation account for the child's age?+

Age adjustments are applied on top of the formula-based recommendation:

  • Under 6: Recommendation is halved and capped at 30 min.
  • 6–9: Applied as-is.
  • 10–12: As-is, with notes on where co-play is advised vs. independent.
  • 13–17: Extended one tier for age-appropriate content — teens benefit from autonomy with guardrails.

Research

How games are reviewed

Who reviews the games?+

Currently, all scores are generated by an AI model (Google Gemini Flash 2.5) working through the full LumiKin rubric — the same rubric documented on this page. The model is given structured game metadata (genre, platform, pricing, monetization flags, ESRB rating) and a set of calibration examples, then asked to score each of the 30+ rubric dimensions and write the parent narratives.

We are not a team of clinical psychologists reviewing games by hand. We are a small project using a transparent, publicly documented rubric applied consistently at scale. The rubric methodology is grounded in peer-reviewed research (cited in each section), but the individual game scores reflect an AI's interpretation of that rubric — not a human expert's. Treat them as a structured starting point, not a clinical assessment.

Human expert review is a goal for the most-played games. If you are a game developer, researcher, or parent who spots a meaningful scoring error, please use the feedback link on any game page.

What data does the AI use to score a game?+
  • Game title, developer, publisher, description, and genre from RAWG
  • Platform availability, ESRB/PEGI rating, Metacritic score
  • Monetization flags: whether the game has microtransactions, loot boxes, a battle pass, or a subscription
  • Multiplayer flags: whether stranger chat is possible and what moderation exists
  • Price and internet requirement
  • The full LumiKin rubric, with calibration examples (Minecraft, Fortnite, Brawl Stars)
How often are scores updated?+
Each score includes a "last reviewed" date. Scores are refreshed when a major update changes monetization mechanics, a new season launches, or community feedback flags a meaningful change. Live-service games are prioritised for more frequent re-scoring. Because the AI re-reads the same rubric each time, scores are consistent across runs — small variations (±2–3 LumiScore points) can occur between model versions.
What if I disagree with a score?+
Every game page has a feedback link. Flagged games are re-examined, and if community feedback reveals a systematic rubric error it feeds into the next methodology version. All rubric weights and thresholds are publicly documented. If a score feels wrong, tell us — the rubric is designed to be transparent and correctable.

Research

Additional context fields

What is the Representation score?+

Each game is assessed on two representation dimensions — gender balance and ethnic & cultural diversity — scored 0–3 each, where higher is better. This is purely informational: it tells you something about the world the game presents, not about how risky the game is. Neither dimension affects the LumiScore or time recommendation.

A score of 0 means characters are all one gender or ethnicity, or rely heavily on stereotypes. A 3 means the game features authentic, diverse representation across both dimensions. Historical games set in genuinely homogeneous contexts are not penalised — context matters.

What is the Ideological content flag?+

Some games carry a political, nationalist, or religious perspective that parents may want to know about before their child plays. The propaganda/ideology level is scored 0–3:

  • 0 — Neutral. No discernible ideological framing (most puzzle, sports, sandbox games).
  • 1 — Mild. Common in historical games with a national perspective. Unlikely to concern most parents.
  • 2 — Notable. Clear political, nationalist, or religious lens. Worth a conversation.
  • 3 — Heavy. Game is primarily a vehicle for ideology or factually distorted content.

This field does not affect the time recommendation or LumiScore. Where level ≥ 1, a short note explains what type of content and where it appears.

Does LumiKin cover VR games?+
Yes. VR games are included in the catalogue alongside standard titles. The rubric handles VR natively — B3.4 (Physical activity) scores highly for motion-based games like Beat Saber, Ring Fit Adventure, and Superhot VR, which is one reason those games often receive extended time recommendations. VR-specific risks (motion sickness, immersion intensity) are captured through the standard content and design risk fields rather than a separate VR category.

Research

Ready to find games?

Browse the catalogue with filters for age, risk level, benefit focus, and more.

Browse gamesGet recommendations