convoluted

Search Engines and Faster Horses

by Rito Ghosh

The Landscape

Complaints with Google has become ubiquitous. Whether it is not being able to find a good result, or being spammed by search-engine optimized result- one can hear complaints about everywhere.

And, many has started competing in this space. Alternative search engines have been spun up. Whether it has big backers like Bing, or small indie-type search engines like Kagi- we see that the search engine space has become crowded in the last few years.

I have personally used Kagi and You.com among search engines- besides Google- to a serious degree. And I seriously liked both of them. I came to know about both of them from Hacker News.

Kagi went paid, and I, the eternal hoarder started saving my Kagi free searches until I practically stopped using it. It's not Kagi's fault, but mine.

I used code.you.com some times, and I liked it. Not because of the result quality, but because of the brilliant UI. I kept using it. My usage of Google gradually kept dropping.

LLMs Happened

Then LLMs, or Large Language Models started to become widely available. After GPT-3, ChatGPT, from OpenAI was made available to the public. I started using it, and began liking it almost immediately.

It hallucinates, but also spews useful stuff in most of the cases.

When you work with code, it not only gives you an answer, it also gives you explanations of each line- like a textbook or a tutor. That's just fabulous in my opinion.

I have used ChatGPT for some of the problems that I faced during work. And it did teach me new library methods that I did not know about earlier, and yet I often come across similar problems setting.

A week ago, I finally decided to set up GitHub Copilot as I already get free access to it. And I started using GitHub Copilot regularly while coding.

I was really charmed by it. Not only did it output correct code each time, but it also quite promptly understood what I was going to do next. It blew my mind. And I do not say that lightly.

*I am searching 80% less Google than before, after I started using Copilot.*

I always checked the correctness before pasting my code, and that is what I am doing with Copilot. So, for me at least, there's no issue with code correctness.

Answer to Search Engines- Aren't Search Engines?

This got me thinking. The way GitHub Copilot has bought down my Google usage- it wasn't made possible by any other search engine at all.

What replaced Google for me- wasn't a search engine at all. It wasn't Brave search or DuckDuckGo or Bing or Kagi- it was something else entirely.

I see a similar pattern in the famous quote by Henry Ford. He famously said-

"If I had asked my customers what they wanted they would have said a faster horse."

What people are looking for may be better search engines. But may be something better can be created.

This is what search engine company founders maybe making a mistake about- they want to replace one search engine with another search engine. Maybe search engines on a whole are ready for disruptions- and not only Google.

Or some uses of search engines can surely be disrupted. Copilot has successfully replaced Google for me for all of my coding needs.

Conclusion

Please think- what if what finally disrupts Google partly or fully isn't another search engine at all? That is a real possibility that we have to take into account.

After all, it's cars that we want, and not faster horses.

Discuss on Hacker News.

Please consider subscribing me on Substack. A Newsletter is about direct connection with an audience, independent of a private, centralized platform. I will NEVER EVER spam you.

#ai #google #search-engines