back to blog

SearchX

SearchX was a people search. You had a name, or a first name and a town, or an address, and you put it in and got a person back.

What it was

#

The idea is simple to say and hard to build: give it a fragment of who someone is and it finds the rest of them. Name and surname. A street address. A phone number. A town. Anything that belonged to a person could be used as a starting point, and the answer was a page, not a link list.

That is the difference between a search engine and a people search. A search engine tells you where the information might be. SearchX told you who the person was, assembled from everywhere at once, and that is the whole product.

Seventeen billion records

#

We collected seventeen billion records. Record is the right word for it: rows, not pages. Each one was a fragment of somebody's life that had already leaked out somewhere and was sitting there waiting to be joined to the rest.

The work was never the collection. The work was the matching. Two records that both belong to the same person almost never look alike, so the entire site lived or died on whether the join was right. When the join is right the site feels impossible, like it knows things. When it is wrong it is worthless.

That is what made it the most filled-in site of its kind. Not the raw number, the coverage: enough data that almost anyone you looked for was already in there, and enough of it to fill in the blanks around them.

The hardest part

#

Scale, and then scale again. Seventeen billion rows is not a database problem you solve once and walk away from. Every query had to come back fast enough that it felt like search and not like a batch job, and every new batch of records pulled at the joins everything else was already sitting on.

It taught me more about data than anything else I have built. Indexing, merging, deduplication, deciding when two people are the same person. None of that is glamorous, and all of it is the product.

Closing it

#

SearchX is closed too. Same reason as slowdown.lol, and it is becoming a pattern: not enough time to keep a site of that size alive. A search index is not a static thing, it is a living one, and it needs feeding and rebuilding constantly.

So there it is. Two sites that both did real numbers and both went quiet, and me at sixteen with the lessons and none of the code still running. That is exactly why I am where I am now, on cheats, on reverse engineering, on things I can keep building with the time I actually have.