Hans Peter Luhn and the Birth of the Hashing Algorithm

The oracle works with bit strings all of the same length. Each bit position represents a single cocktail.

Each ingredient card is a bit string. Each bit is 1 (opaque) if the ingredient is needed for the cocktail at that bit position.

You take all the ingredients you have and discard their bitstrings ("turn down their cards"). Then you or together the remaining bitstrings of all of the ingredients you don't have (look through their overlapping cards). In other words, not having an ingredient forbids certain cocktails. Any bit positions that remain zero (transparent all the way through to the key card on the back showing the names of the cocktails at each position) are cocktails that you have enough ingredients to make.

Using opacity to represent bitwise "or" is pretty clever. Turning down the ingredient cards you don't have is really clever. I kind of want to make one of these now.

Back in 1953, IBM was investigating disk storage for record retrieval in large volumes of data. Seek/read time on these machines was on the order of a second, so a 10 level binary search could take around 10 seconds of total time. Unfortunately for IBM, this wasn't an improvement over the best human-driven manual lookup processes... which is an issue if you're trying to sell/lease expensive computer hardware.

With this as motivation, Luhn then essentially invented what we now would know as a hash table, which was key to the 1956 RAMAC release.

Perfectly obvious in hindsight but I never made the connection.

Ctrl+F the article for "hash". What Luhn invented was a checksum, one that is apparently still used today in IMEIs (among other things), but the article says nothing about him furthering checksums into hash functions (and there is quite a distinction). Sure, a hash function can hardly be invented without going through the checksum stage, but it's like saying the person who discovered there's oil in the ground invented cars just because he laid the foundation.