How We Catalog Board Games by How They Feel

The box tells you what’s inside a game. It rarely tells you what an evening with it will feel like. That gap is why we built The Ludarium Card Catalogue: a library-style system that files every game by its form and describes it by six dimensions of feel.

You know the moment. Four friends, one shelf, twenty minutes of debate. Someone suggests the game with the best ratings. An hour later, one person is quietly miserable, two are lost in the rules, and nobody laughed.

The problem isn’t the people or the games. It’s the language. “Strategy,” “family” and “card game” describe what’s in the box. They say nothing about whether the table will be loud or silent, whether you’ll fight each other or work together, or whether a new player will be lost for an hour.

At The Ludarium we believe the game is the reason and the people are the point. So we catalog games the way a librarian would, but with one question in mind: what will this game do to the people around the table?

Where this idea comes from

Our system builds directly on work by Tabletop Library, a membership board game club in Berkeley, California. In July 2026 their team published a detailed write-up of their game recommender, and it is one of the best things written about choosing games for a group. Read it.

They started by asking AI chatbots what to play. Across 600 test recommendations, the chatbots returned only 61 different games, leaning hard on the same famous titles. They also ignored basics, like suggesting a two-player game for a group of four.

Their fix had three parts:

• A call-number system for games. Their Tabletop Library Classification System shelves each game by its primary form, so drafting games, campaign games and social deduction games live in different sections.
• Five dimensions of feel. Every game is rated on theme, randomness, player interaction, learning curve and tempo, each with three levels.
• Member taste profiles. Ratings plus “can teach” and “want to learn” lists become a profile in that same vocabulary. A one-star “Not For Me” rating acts as a veto.

Their biggest lesson: the model was never the bottleneck. The vocabulary was. A system can only weigh what it has words for. We agree, and we think the vocabulary can go further.

The shelves: call numbers for games

Every game gets a call number, just like a library book. Ten shelves, numbered by the hundreds, group games by what you actually do when you play. Each shelf has subsections, and each card ends with a three-letter title code and the year.

So Wingspan is 520 WIN 2019: shelf 500 (Engines & Tableaus), subsection 520 (Tableau engines). Euchre, a Midwest kitchen-table classic, is 310 EUC: Classic Cards, trick-taking. Classics have no year.

Call no.	Shelf	Subsections	Example
000	Party & Dexterity	010 Word · 020 Guessing · 030 Drawing · 040 Storytelling · 050 Dexterity	Codenames
100	Deduction & Hidden Roles	110 Hidden roles · 120 Bluffing & clue-giving · 130 Logic · 140 Co-op deduction	The Resistance: Avalon
200	Cooperative	210 Crisis management · 220 Limited communication · 230 Heavy co-op · 240 Two-player co-op	Pandemic
300	Classic Cards	310 Trick-taking · 320 Pegging & melding · 330 Shedding & climbing	Euchre
400	Drafting & Collecting	410 Card drafting · 420 Sets & routes	Ticket to Ride
500	Engines & Tableaus	510 Deck-building · 520 Tableau · 530 Economic · 540 Worker placement	Wingspan
600	Head-to-Head	610 Dice brawls · 620 Asymmetric conflict · 630 Area control · 650 Racing	Root
700	Trade & Negotiation	710 Trading · 720 Negotiation & bribes · 730 Auctions	Catan
800	Abstract & Puzzle	810 Tile & pattern · 820 Spatial puzzles · 830 Classic abstracts	Azul
900	Story & Campaign	910 Dungeon campaigns · 920 Mystery & narrative · 930 Living card campaigns	Sherlock Holmes Consulting Detective

A game goes on one shelf only, chosen by its main activity. Ticket to Ride has a map, but most of your turns are collecting sets of cards, so it lives at 420.

The six dimensions of feel

The shelf tells you what you’ll do. The six dimensions tell you how it will feel. Each one has three levels, so any game can be described in six quick answers. We kept Tabletop Library’s five dimensions and added a sixth: Table Talk.

Dimension	The question it answers	Level 1	Level 2	Level 3
Theme	How much does the world matter?	Abstract	Themed	Immersive
Chance	Who decides the winner, dice or decisions?	Luck-driven	Tactical	Skill-driven
Interaction	How much do other players change your plans?	Solo-ish	Indirect	Direct
Learning	How long until a new player gets it?	Intuitive	Moderate	Heavy
Tempo	How much brain does a turn take?	Light	Thoughtful	Intense
Table Talk	How much talking and laughing does it create?	Quiet	Chatty	Raucous

Why Table Talk matters most to us. Two games can match on every other dimension and still create completely different evenings. Hanabi and Codenames are both easy to learn and both built on giving clues. One is played in near silence. The other has the whole table shouting. If your goal is togethering, how much a game gets people talking is the single most useful thing to know.

A few notes on reading the levels:

• Interaction counts cooperation. In a co-op game every decision changes your teammates’ plans, so Pandemic rates Direct.
• Quiet isn’t bad. Sky Team is nearly silent by design, and that tension is the fun. Quiet games make great date nights.
• Levels describe the game, not quality. A Heavy, Intense, Immersive game isn’t better than a Light one. It’s different, and right for a different night.

Six ways we improved the system

Tabletop Library was candid about what their recommender still gets wrong. We designed the Card Catalogue to close those gaps from day one, and added a few ideas of our own.

1. Table Talk, a sixth dimension. As above. It measures the thing our mission cares about: whether people connect.
2. Sweet-spot player counts. A box that says 1 to 5 players hides the truth that many games shine at one or two counts. Every card lists a “best at” range, so the catalogue knows a game that seats five may sing at three.
3. Real play times. Box times are guesses. You log how long each play actually took, how many people played, and whether someone was learning. Over time each game gets a real run time for your group size, not one number printed on a box.
4. Togethering outcomes. After each play, you answer two quick questions: would you play again with this group, and how loud was the table? That closes the biggest gap in most recommenders, which only guess whether a pick looks right. We measure whether the evening actually worked.
5. Discovery picks. Recommenders tend to serve twelve versions of the same game. Every set of picks we make includes one Discovery pick from a shelf you rarely visit, chosen to still fit your profile.
6. Occasions. Game nights have contexts the dimensions miss. Every card is tagged for occasions like Date night, Family, New friends, Team building, Game group and Two-player.

We also keep what Tabletop Library got right: hard rules first, vetoes respected, and teachers matched with learners.

Your Play Profile

Every game you rate or log teaches the catalog what you like, in the same six-dimension language. After about three games, your Play Profile starts making picks.

Here’s how each signal counts toward your profile:

What you do	Weight
Rate a game 5 stars	+2
Rate a game 4 stars	+1
Rate a game 3 stars	0
Rate a game 2 stars	−1
Rate a game 1 star	−2
Mark it “Can teach”	+1
Mark it “Want to learn”	+0.5
Log a play and say you’d play again	+0.5
Log a play and say probably not	−0.5
Mark it “Not for me”	Veto: never recommended

Each weight is added to the levels that game holds on all six dimensions. If you love Codenames, Wavelength and Just One, your profile lights up Chatty-to-Raucous on Table Talk and Intuitive on Learning. Picks then favor unplayed games that sit on those same levels. A game that misses your favorite level counts against itself, and Table Talk carries 1.5 times the weight of the other five dimensions, because it’s the one most tied to whether people connect.

Picking for a group. When several people play together, each person’s fit is scored separately and then combined with a geometric mean instead of an average. That way one unhappy player pulls a game down hard. Tabletop Library’s example makes the case: three players at 0.9 and one at 0.2 average to 0.73, which looks fine. The geometric mean is about 0.62, which matches the fourth player’s face.

Two more group rules come first. Any “Not for me” at the table removes the game. A Heavy game only gets picked when someone at the table can teach it.

Use it at your next game night

You don’t need an app to start. Before your next game night, ask the group three questions:

1. How loud do we want tonight to be? That’s Table Talk. Quiet, Chatty or Raucous.
2. Do we want to work together or against each other? That’s Interaction, and it rules out more games than anything else.
3. Does anyone need to learn it? If yes, stay Intuitive, or pick a game someone can teach.

Three answers usually narrow a shelf of forty games to four or five good ones.

FAQ

Why only three levels per dimension? Three is enough to separate games that feel different, and few enough that anyone can rate a game in under a minute. Finer scales invite arguments about 6 versus 7.

Who decides a game’s levels? We file every card, and we revise cards as we learn. Your own play logs show how a game really runs at your table: if your group logs a “Chatty” game as Raucous every time, you’ll see it.

What if a game fits two shelves? It goes where you spend most of your turns. The six dimensions capture the rest.

Can I add a game you haven’t cataloged? Yes. In the Card Catalogue you can file your own card, with a shelf and all six dimensions.

Start your Play Profile

The Card Catalogue is free and open to everyone, with no account. Rate a few games you know, log your plays, and watch your Play Profile take shape. Your cards stay on your own device. Then bring your friends to the table: share a table code, add guests, and let the catalogue find a game everyone will enjoy.

Open the Card Catalogue

Never stop playing.

Sources

• Nabeel Hyatt, Making a Better Board Game Recommender, Tabletop Library, July 11, 2026.
• Carlo Bianchini and Paolo Munini, ClaG: a faceted classification of plays and games for game libraries, University of Pavia.