2
1 Comment

I stopped trusting search results. So I started building my own search engine.

I've noticed something frustrating about search lately.

The internet keeps getting bigger...

but finding the right answer doesn't always feel easier.

A lot of searches now lead to:
- SEO-filled articles
- pages written for ranking instead of helping
- outdated content
- results that technically match but miss the intent

The problem isn't that search engines can't find pages.

The problem is finding the page you actually wanted.

That got me thinking:

What would a search engine look like if it was built from scratch today?

Not a wrapper around another search API.

Not a browser feature.

A real search engine with its own crawler, index, and ranking system.

So I started building Slick.

The idea is simple:

Give users more control over how they search.

I'm experimenting with things like:
- independent web crawling
- semantic search
- customizable ranking
- privacy-focused design
- letting users influence their own results

The interesting part is that search has a lot more tradeoffs than I expected.

Improving one thing can hurt another.

More personalization can improve relevance, but can also hide new information.

More AI understanding can help with intent, but sometimes exact keyword matching is still better.

A bigger index doesn't automatically mean better results.

I'm still early, and I don't think building a search engine is something you "finish".

It's a constant process of improving relevance.

I'm curious...

When you use a search engine today, what is the biggest thing that annoys you?
Is it:
- too many ads?
- SEO spam?
- bad relevance?
- privacy concerns?
- something else?

I'm building Slick around these problems and would genuinely love feedback from people who search the web every day.

https://slicksearchhq.com

posted toAvatar for product Slick
Slick
  1. 1
    Respect for building your own crawler and index. I went the opposite direction in 1998 — instead of crawling the web, I proposed making the web push changes to the index. Built a Windows DDK prototype: filesystem driver detects local changes, extracts text/links, compresses the payload, pushes only deltas to the search engine. No crawler needed. Sent the proposal to Microsoft from rural Russia. Their reply: five lines. "Does not fit our current business needs." Yandex thought I was building a data theft tool. The startup died at the US embassy. Twenty years later the industry arrived at the same architecture via PubSubHubbub and IndexNow. The idea wasn't wrong — the timing was. I still have the letter.