They literally don’t make Boggle like they used to.

The words you can spell in a Boggle game depend on the letters on 16 dice rolled into a four-by-four grid. The letters on the dice are carefully selected — like in Scrabble, common letters like E and A appear more often than uncommon ones like V or the QU ligature. In fact, depending on when your English-language game of Boggle was manufactured, you have one of two sets of dice.

• The “Classic Boggle” set, produced from 1976 to 1986.
• The “New Boggle” set, produced from 1987 to today.

There’s scant info about the 1986 redesign online — none from the designers themselves — but notably New Boggle makes letters F and K mutually exclusive, rendering FUCK unplayable.¹ This blog post by ‘Bananagrammer’ describes the differences between the dice sets in detail.

Dice set	Average # of words	Average Boggle score	Average length of longest word
New Boggle	~104	~150	6.8
Classic Boggle	~93	~128	6.6

[…] Of course, these calculations only confirmed what I already knew: the older version of the game is harder and is the one for me. I bought a copy of classic Boggle from eBay. The dice are made of wood rather than plastic. The timer has sand in it and doesn’t make some noise to tell me when time is up. Succinctly, I think it is skookum.

I think Bananagrammer got this wrong.

The fun in Boggle isn’t really in the score — it’s in the words themselves: quality over quantity, style points, that kind of thing. A good evening of New Boggle yields higher per-board word counts, but how does it affect the diversity of play?

Dictionaries

A set of Boggle dice sets certain hard constraints on word formation. Each dice set contains a single Z; kiss JAZZ goodbye. Moreover, letters that share die are mutually constraining, since only one side of a given die affects any given game. Classic Boggle’s only QU and only J are on the same die, so JERQUE is impossible (just like FUCK in New Boggle). You can speak of a given die set’s “dictionary;” start with a standard dictionary (CSW19) and filter out the categorically unplayable words.

Classic Boggle has a substantially larger dictionary: 276,339 playable words, 18,517 more than New Boggle. A full 6.7% of the Classic dictionary is unplayable today! Only 134 words are uniquely playable in New Boggle.

At this juncture, it seems like Bananagrammer has a point: New Boggle gives you more words per game on average, in a smaller (dumbed-down??) dictionary.

Same-games

Let’s return to the question: diversity of play. A comparison between Classic and New Boggle should consider typical boards, not theoretical maximum dictionaries. Vanishingly few games of Classic Boggle admit 15-letter UNDEREMPLOYMENT; its 1987 exclusion barely impacts the game’s diversity in practice. A good set of dice don’t just admit a large theoretical dictionary of words, but also provide difference game-to-game. A good set of dice yields Boggle games that aren’t samey.

Suppose you play 10 games of Classic or New Boggle in an evening. How many unique words can you actually find across those games?

By this criterion, there really isn’t a reason to prefer Classic Boggle over New Boggle. While it has a substantially larger dictionary, Classic Boggle offers fewer words per game — Bananagrammer’s preference, but not mine — and offers no meaningful difference in the variety of words actually encountered game-to-game.²

Play and Experimentation

The Boggle solver I wrote for these experiments ships with a terminal interface for solitaire games with my house rules.³ You can play a hosted game by running ssh boggle.fly.dev, or install the game from source:

go install github.com/lukasschwab/boggle@latest

Games look like this:

$ boggle
0/50 • 2m41s • bGNsbmFuYnVhYXBoc3Jsbg==
┌──────────────┐  
│  l  c  l  n  │  
│  a  n  b  u  │  
│  a  a  p  h  │  
│  s  r  l  n  │  
└──────────────┘  
> club|
[enter] submit word • [ctrl+c] quit

You can play with either dice set — see boggle --help. See the experiments branch (especially the cmd directories) to reproduce the results in this post.

If you’re interested in Boggle solvers, implement one yourself! This was a good data-structures exercise, and tweaking the game’s parameters . Curious about optimization problems across Boggle boards, e.g. a search for the Boggle board with the greatest number of findable words? Read Dan Vanderkam’s “What up with Boggle” series.