Pray, Mr. Babbage, what about search?

I’ve been thinking about this comment on and off since reading it months ago:

“The entire problem of search is that the user has the wrong data and wants to use it to receive the correct data.”

– User devmor on Hacker News (link)

It’s directly in contrast to this 1864 quote from Charles Babbage, which reminds me of certain personal experiences:

On two occasions I have been asked, 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?'

I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.

– Charles Babbage, Passages from the Life of a Philosopher (1864)

Exact-matching search is simple: given a search query, find documents that contain that exact query.

But in “smarter” searching, we might broaden the criteria to find other things you might be searching for. Maybe we detect and suggest corrections to common typos:

Or instead of being strict about “meow”, we might accept variants like “meows” or “meowing” (but not unrelated substring-matching words, like “homeowner”). 

In natural language processing, “stemming” converts word variants into their common base word.

And if we wanted to really meet our users where they are, we can slam that entire query into an LLM (or even further, rewrite their prompt for them before interpreting it). 

When it works, it’s magical. But when it’s not working, maybe because it’s tuned to the quirks of its creator or a chatbot is just the wrong interface, it makes me wonder…

Pray, Mr. Babbage, if I put into the machine the right figures, but it decides I must have meant something else, will the right answers come out?