a7i7 / smriti

Memorize your seed phrases

Smriti

Smriti is an approach to help you transform your 12 or 24 word seed phrases into memorizable ones.

There are 2048 words within the BIP-39 wordlist.

This gives an entropy of $\\2048^{12}=2^{132}$ and $\\2048^{24}=2^{264}$ for the 12 and 24 word seed phrase respectively.

Memorization

While it is difficult for someone to memorize a list of 12 / 24 random words in order,
it is relatively easy for someone to memorize a number of different classes of things.

e.g.

Remember a Movie : Les Miserables

Remember a Fruit : Kiwi

Remember a Bird : Humming bird

Remember a song : The Amen Corner

and so on

The list of such classes (Movie, Fruit, Bird etc.) are ordered which reduces a
burden of the end user.

They only need to remember the names of objects from each class
but has no need to maintain the order in their head.

Construction

So the math works out to

$\\{N(c_{1})}*{N(c_{2})}*{N(c_{3})}*...*{N(c_{m})}=2^{132}$

$\\{N(c_{1})}*{N(c_{2})}*{N(c_{3})}*...*{N(c_{m})}=2^{264}$

for the 12 and 24 word seed phrase respectively

where we have m different classes of things to memorize, with

$\\{N(c_{i})}$ is the number objects in class i

If we assume each class to have the same number of objects,

$\\{N(c_{1})}={N(c_{2})}={N(c_{3})}=...={N(c_{m})}$

this table lists how many different classes we will need

Number of classes (m)	for 12 word seed phrase	for 24 word seed phrase
10	9411	88550677
11	4097	16777217
12	2048	4194304
13	1140	1297851
14	690	474861
15	446	198669
16	305	92682
17	218	47296
18	162	26008
19	124	15232
20	98	9411
21	79	6087
22	65	4097
23	54	2853
24	46	2048
25	39	1510
26	34	1140
27	30	878
28	27	690
29	24	551
30	22	446
31	20	367
32	18	305
33	16	257
34	15	218
35	14	187

In reality not all the classes will have the same number of objects but the table gives us a rough idea of how easy it will be to create a memorizable set of classes.

For example, it is very easy to create 30 classes with 446 elements each.

However memorizing 30 different objects is very hard for the end user.

On the other hand memorizing only 10 elements is easy for the end user,
but constructing 10 classes with 80 million + elements is not easy.

e.g. We were able to compile a list of international movies of only 45000+ elements.

Suggested classes

Right now I am constructing the list of classes and elements within each that will make your seed phrases memorizable. Here are some suggestions

• Movies
• Birds
• Historical Figures
• Song / Music

Contributions are welcome as pull requests or discussions in Github issues.

Feel free to reach out to me at my website