hfukada

Background

Simplifying somewhat, each word of English is composed of a string of sounds, called "phones"/"phonemes".  For example, "cat" has the phones /k/, /a/, /t/.

Some phones are more "confusable" with other phones--it's easier to mishear /m/ (as in "ram") as /n/ (as in "ran") than /t/ (as in "rat").  I happen to have a phone-to-phone confusion matrix, which tells me the probability of confusing any phone as any other phone.

If I want to calculate how easy it is to confuse one string of phones with another of equal length (e.g., hearing "ram" as "rat"), I simply multiply the probability of hearing each phone in the first string as the corresponding phone in the second string.  E.g., P(/r/|/r/) * P(/a/|/a/) * P(/t/|/m/).  Like I said, I get these probabilities from my confusion matrix.

hfukada

" Settings
"set nosmoothscroll
set noautofocus " The opposite of autofocus; this setting stops
                " sites from focusing on an input box when they load
set typelinkhints
let searchlimit = 20
let scrollstep = 50
let barposition = "top"
let hintcharacters = "asdfjkl"
let completionengines = ["google", "amazon"]

hfukada

def gen_nonce():
    """Generates a random string of bytes, base64 encoded"""
    length = 64
    string = base64.b64encode(os.urandom(length), altchars=b'-_')
    b64len = 4 * min(length, 3)
    if length % 3 == 1:
        b64len += 2
    elif length % 3 == 2:
        b64len += 3
    return string[0:b64len].decode()

hfukada

### Keybase proof

I hereby claim:

  * I am hfukada on github.
  * I am hiroshf (https://keybase.io/hiroshf) on keybase.
  * I have a public key whose fingerprint is 2E52 8A40 D052 B98C 3AE7  E119 9DE5 5147 A170 3CCE

To claim this, I am signing this object: