This is very interesting. I work on the Expert Systems side of things, having (back in 1984) created an Expert System, named XTRAN, that knows more than 40 computer languages, including assemblers, 3GLs, 4GLs, DSLs, scripting languages, and data base languages. Its rules language can be used to capture and therefore automate coding skills, including assessment / analysis, transformation / re-engineering, translation, and code creation, as well as automating the manipulation of data and text.
It's also interesting because, as a software entrepreneur from 1977, my first commercial product was a sorting utility for the DEC PDP-11 minicomputer. Its unique feature was a high-performance n-way merging algorithm created by a colleague of mine.
And it's ineresting, thirdly, because back around 1970 I was handed the care and feeding of an early DBMS that was based on hashing. And earlier, in the mid-1960s, I was given the task of storing a large insurance policy master file (more than a million punched cards) on drum (!) storage for random access. As a result, I researched early hashing algorithms, with the idea of hashing on policy numbers. One thing I learned was that further complicating an already complex hashing algorithm typically worsens its collision behavior. I wonder if AlphaDev has figured that out. :)
Nowadays I'm using hashing to implement associative arrays in XTRAN. Some things never go out of style. :)
This is very interesting. I work on the Expert Systems side of things, having (back in 1984) created an Expert System, named XTRAN, that knows more than 40 computer languages, including assemblers, 3GLs, 4GLs, DSLs, scripting languages, and data base languages. Its rules language can be used to capture and therefore automate coding skills, including assessment / analysis, transformation / re-engineering, translation, and code creation, as well as automating the manipulation of data and text.
It's also interesting because, as a software entrepreneur from 1977, my first commercial product was a sorting utility for the DEC PDP-11 minicomputer. Its unique feature was a high-performance n-way merging algorithm created by a colleague of mine.
And it's ineresting, thirdly, because back around 1970 I was handed the care and feeding of an early DBMS that was based on hashing. And earlier, in the mid-1960s, I was given the task of storing a large insurance policy master file (more than a million punched cards) on drum (!) storage for random access. As a result, I researched early hashing algorithms, with the idea of hashing on policy numbers. One thing I learned was that further complicating an already complex hashing algorithm typically worsens its collision behavior. I wonder if AlphaDev has figured that out. :)
Nowadays I'm using hashing to implement associative arrays in XTRAN. Some things never go out of style. :)