A dictionary attack (辞書攻撃) guesses a password by trying, one after another, a list of words and strings that people tend to choose as passwords.
The attacker prepares only likely candidates such as names, English words and common number sequences, so a password can sometimes fall after only a few attempts. Strings that are not on the list never match, so choosing a random string over a meaningful word is an effective defense.
AnalogyIt is like a burglar hunting for a spare key who checks only the usual hiding spots, such as under the flowerpot or behind the mailbox, and ignores everywhere else.
How to rememberThe “dictionary” here is not a language dictionary but the attacker's own word book of commonly used passwords.
Exam points
ExamA dictionary attack (辞書攻撃) works out passwords from a word list, which sets it apart from a keylogger (キーロガー) that records keystrokes on the user's PC and steals them.
Mix-upA dictionary attack narrows its candidates to commonly used words, whereas a brute-force attack (ブルートフォース攻撃) tries every combination without narrowing.
A brute-force attack (ブルートフォース攻撃) is an exhaustive search (総当たり) that works out a password by trying every combination of the usable characters in turn.
Length (36 characters: lowercase letters and digits)
Maximum attempts
1 character
36
2 characters
1,296 (36 × 36)
3 characters
46,656 (36 × 36 × 36)
Each extra character
36 times as many
Because it does not narrow its candidates, it reaches the correct password eventually if given enough time. With n usable characters and a length of k, the number of combinations is n to the power k (k factors of n multiplied together), so the attempts rise steeply as the length or the character set grows. A mechanism that temporarily disables an ID after a set number of consecutive wrong entries is called account lockout (アカウントロック).
AnalogyIt is like opening a three-digit combination lock by turning it through every number from 000 to 999.
How to remember“Brute force” means sheer force: the attack wins by numbers rather than by clever guessing.
Exam points
ExamWith n usable characters, each extra character multiplies the maximum number of attempts by n, and two extra characters multiply it by n squared.
ExamAccount lockout (アカウントロック) works well against a brute-force attack (ブルートフォース攻撃) because it cuts off a flood of attempts against a single ID partway through.
A password list attack (パスワードリスト攻撃) uses a list of ID–password pairs leaked elsewhere to try logging in to a different site.
IDs and passwords leak from Site X
The attacker obtains the list of pairs
The same pairs are entered on Site Y
Accounts that use the same pair are taken over
Only users who used the same pair on the other site are broken into at this step.
4 / 4
The attacker does no guessing and simply enters pairs that were actually in use. Damage spreads because a pair leaked from one site also works on any other site where the same pair is registered.
AnalogyIt is like someone who opens their house, car and locker with one key: if they drop it, whoever picks it up can open all three.
Exam points
ExamThe damage from a password list attack (パスワードリスト攻撃) comes from reusing (使い回す) the same ID and password across several sites.
ExamAccount lockout (アカウントロック), which stops an ID after repeated errors, does little against a password list attack, since that attack tries leaked pairs only a few times per ID.
Credential stuffing (クレデンシャルスタッフィング) tries leaked credentials (ID–password pairs) one after another on the logins of other sites, and it names the same technique as a password list attack.
Whichever name is used, the attack relies on real pairs leaked somewhere else. Tools often enter large numbers of pairs automatically, so a single leak can lead to unauthorized logins on many sites.
How to rememberCredential means login information and stuffing means cramming in: the name pictures leaked credentials being crammed into login boxes one after another.
Exam points
ExamCredential stuffing (クレデンシャルスタッフィング) and a password list attack (パスワードリスト攻撃) both mean breaking into a target site with ID–password pairs leaked elsewhere.