# Longest word worth at most a million

An old and popular puzzle, recently revived on Twitter by Alex Bellos, Chris Smith, and others, asks to take the first 26 primes, relate each to letters of the alphabet (A = 2, B = 3, C = 5, ..., Y = 97, Z = 101), and use this code to find the (legitimate) English word that comes closest to a million, where the word's numeric value is given by the product of its letters. CAB is thus valued 5x2x3=30.

With the same code, what is the longest English word valued at most a million?

• Well, an upper bound is $19$ letters as $2^{19} = 524288$.
– Jens
Nov 10, 2019 at 2:30
• might be useful: answers.com/Q/…
– JMP
Nov 10, 2019 at 3:04
• What counts as an English word, exactly?
– Deusovi
Nov 10, 2019 at 3:06
• @Deusovi Say a word admissible to Scrabble players. Nov 10, 2019 at 3:09
• Vaguely similar to this one I asked
– smci
Nov 11, 2019 at 6:29

## 6 Answers

Closest to one million:

COLIC = 999925

This is the sixth-largest number less than or equal to one million with no prime factors greater than 101.

As for longer words:

Without much effort, I have found several seven-letter words with values less than one million. In particular, the least among them is CABBAGE = 33660. I expect their is some eight-letter word to be found.

• I came to the same result. Nov 10, 2019 at 18:53

Using the scrabble dictionary, and a short python program:

Several 8-letter words, I think the best one is
CABBAGED: 235620 "To steal, pilfer"
But the 8-letter one closest to 1,000,000 is:
ANABAENA: 976272 "A genus of filamentous cyanobacteria that exist as plankton"
Other 8-letter words: cabbalah: 253080, cabbalas: 892440, dahabeah: 667128.

• Note that 'closest to one million' and 'longest not more than one million' were two separate queries Nov 10, 2019 at 4:25
• @Daniel Mathias Ah, the "closest to one million" means of the 8-letter words. Edited answer for clarity. Nov 10, 2019 at 4:28
• Can you please share the Python code?
– user62885
Nov 11, 2019 at 0:27
• Uhr... I thought the solution was the old smiles joke, with two s's removed... Nov 11, 2019 at 1:03
• @Nyssa Essentially the same as CDJB's answer. Nov 11, 2019 at 19:15

I used an english wordlist from here, and a short python program to yield:

Longest possible word under 10**6:

calabaza (8 characters) - 896880

Closest to 10**6:

colic - 999925

Code used:

import numpy as np
SCORER = {'a': 2,
'b': 3,
'c': 5,
'd': 7,
'e': 11,
'f': 13,
'g': 17,
'h': 19,
'i': 23,
'j': 29,
'k': 31,
'l': 37,
'm': 41,
'n': 43,
'o': 47,
'p': 53,
'q': 59,
'r': 61,
's': 67,
't': 71,
'u': 73,
'v': 79,
'w': 83,
'x': 89,
'y': 97,
'z': 101}

def load_words():
with open('words_alpha.txt') as word_file:
valid_words = set(word_file.read().split())
return valid_words

def getScore(word):
return np.prod([SCORER[char] for char in word], dtype=np.uint64)

english_words = load_words()

longest = ''
closest = ('', 0)

for word in english_words:
score = getScore(word)
if score < 10**6:
if len(word) > len(longest):
longest = word
if 10**6 - closest[1] > 10**6 - score:
closest = (word, score)

print(longest, getScore(longest))
print(closest)

To get the ball rolling, I've got

CABBAGE
Length: 7
Value: 33,660

Note: this does not answer the question asked; it answers the question of finding the best possible value at most a million, not the longest length.

Here's a Mathematica notebook (via Wolfram Programming Lab) that I used to check for better solutions: https://www.wolframcloud.com/obj/solly.ucko/Published/Longest_word_worth_at_most_a_million.nb

In[1]:= letterMapping = AssociationThread[Prime[Range[26]], Alphabet[ ]]
Out[1]= <|2 → a, 3 → b, 5 → c, 7 → d, 11 → e, 13 → f, 17 → g, 19 → h,
23 → i, 29 → j, 31 → k, 37 → l, 41 → m, 43 → n, 47 → o, 53 → p, 59 → q,
61 → r, 67 → s, 71 → t, 73 → u, 79 → v, 83 → w, 89 → x, 97 → y, 101 → z|>
In[2]:= factorList[n_] := Flatten[Map[Apply[ConstantArray], FactorInteger[n], 1]]
In[3]:= f[n_] := Map[letterMapping, factorList[n]]
In[4]:= factorList[1 000 000]
Out[4]= {2, 2, 2, 2, 2, 2, 5, 5, 5, 5, 5, 5}
In[5]:= Select[Table[f[n], {n, 1 000 000, 1 000 000 - 75, -1}], Not@* AnyTrue[MissingQ]]
Out[5]= { {a, a, a, a, a, a, c, c, c, c, c, c}, {b, b, b, d, e, f, l},
{f, f, r, y}, {j, j, j, m}, {a, a, a, a, a, a, a, a, a, b, b, d, k}, {c, c, i, l, o} }

So unless a permutation of {a, a, a, a, a, a, c, c, c, c, c, c}, {b, b, b, d, e, f, l}, {f, f, r, y}, {j, j, j, m}, or {a, a, a, a, a, a, a, a, a, b, b, d, k} is a valid word (unlikely), Daniel Mathias's answer of

COLIC

is optimal.

• @JiK Oh, oops... I'll work on fixing it at some point if/when I get a chance. Nov 11, 2019 at 14:10

I wrote a program to check a list of words from

In addition to

ANABAENA,

which others have found, here are the words of each length with the lowest scores:

A (2) AA (4) ABA or BAA (12) ABBA or BABA (36) ABACA (120) BACCAE (3 300) CABBALA (13 320) CABBAGED (235 620) ABDICABLE (11 794 860) DEADHEADED (242 875 556) ABRACADABRA (37 507 680) ABRACADABRAS (2 513 014 560)

No words over 12 letters long are listed, possibly at least in part because a Scrabble board is 15x15.

Interestingly, there is an 11-length word with a lower score than any 10-length word.