Browser developer discusses his expectations of search engines | Hacker News Reader

Browser developer discusses his expectations of search engines

stoutner.com

43 pointsColinHayhurst17 comments

I don't think you have to invest billion and build a general search engine to compete with Google. Just look at where Google makes its money: commercial searches. The searches for products and services are where you see some of the most problematic results and also where Google probably makes the bulk of their search revenue. Nobody is going to bid an ad for the search "who was the 15th president of the US" but they do bid for "apple airpods" or "airpods alternatives."

So even if you just take one of these commercial search verticals, such as product searches, you can build a compelling product that effectively takes a massive chunk of Google profits.

So even if you just take one of these commercial search verticals, such as product searches, you can build a compelling product that effectively takes a massive chunk of Google profits.

That will only work if users are willing to go to Acme Search to search for products and Google for everything else. If they're not, and they just stick with Google for everything until they can switch to a different search engine entirely, then you have to be better than Google for everything regardless of whether or not you can sell ads on the searches.

If you can't get users to use your search engine then it doesn't matter how good it is.

To be fair, this is improving - chromium browsers let you specify a particular search engine with a keyword, even without changing your default. It almost certainly isn't mainstream enough and not enough is known about the competitors, but some of the groundwork has been laid

Does Chrome still spam your search provider list with every site you visit that opts in?

People do this all the time. I don’t search Google when I’m looking for restaurants near me… I just use Door Dash, Uber Eats, or Google Maps (yes I know it’s owned by Google but it’s a separate product / UI from search) directly.

For restaurant reviews, I go straight to Yelp.

For used products, casual encounters, craigslist or ebay.

For products I sometimes do google and sometimes go straight to Amazon.

For friends, facebook. In fact this site is an example of one that is actually completely non-indexed by Google.

For the certain types of products rtings.com is known for I go straight there… Of course this one only applies if you know about rtings and it’s relatively niche still.

I think my usage pattern is not unique. Many people skip Google for certain verticals already, and often Google’s search coverage is not as good as the other service’s native faceted search. Google’s search is the catch-all last resort for me if I don’t know of a better alternative in the vertical I’m interested in.

I think this is more location (country, continent, city) dependent than you think, and you are probably more platform-savvy (being, that you are on HN), than the average person. People generally don't go to ten platforms, if they can get 90% of the way, on a single one.

Imagine a search engine that only indexes quality review sites (like Rtings), and has a comprehensive database across all of commerce. It would become super popular, and would be easy because there are very few quality review sites.

The hardest part is not selling out and becoming yet another shady bad review site/indexer.

I didn't know about Rtings and they look interesting. Who publishes them, do you know? It's not discussed on their About page. Also, in some ways they seem similar to Consumer Reports. Have you seen CR and how would you compare them? Thanks!

Sure the more savvy you are the more “better than google” services you know. But let’s be honest DoorDash, Uber, Yelp are all pretty much household names.

Being able to do the right google search also requires some saviness so probably a person who doesn’t know Yelp exists also doesn’t know how to internet.

> Nobody is going to bid an ad for the search "who was the 15th president of the US" but they do bid for "apple airpods" or "airpods alternatives."

They do. SEO spammers have been cluttering results for simple queries over the years, and it only seems to get worse.

The typical pattern is: lengthy paragraphs that introduce simplistic questions and definitions with ads scattered across them. Often the answer is burried deep into the “article”.

Similarly there are plenty of spam sites that aggregate github issues, SO QA, and official forums and docs. Again sprinkled with dozens of ads and typically no source cited.

Search engines don’t seem to be able to keep up and their utility degrades. It might be a combination of conflict of interest and inability to deal with the sheer complexity and volume.

I’m talking about ads, but also if you actually do the search I mentioned there are not really SEO spammers nor any ad results.

Spot on, and that's why Amazon's ad business is booming.

Price comparison sites used to be very popular. But product search is extremely hard to do well. Users want to compare products on their merits, not based on how much the seller is bidding on PPC. And not just product features but also seller features: logistics, delivery, reliability, reviews etc.

I think Amazon has largely stolen that crown from Google already. Even though their search isn't great either.

Where Google seems to make money is the commercial search for products and services NOT sold on Amazon: SaaS, insurance, medical, legal etc.

The fundamental problem is that the PPC model doesn't align incentives: the search engine doesn't optimise for relevancy but for revenue. We need a different model.

It seems to me that Amazon is also losing out on some higher end products these days too.

I'm convinced Apple will announce something soon. Applebot is getting busier and busiers and it's crawling my server at about 1 rq/s now.

There's definitely something to this as Apple's been building a homegrown search platform for several years. My guess is that whatever they announce at WWDC will still be tied to Siri — for now.

https://appleinsider.com/articles/22/06/02/rumor-that-apple-...

As far as I know search ranking on Google is largely driven by AI these days, which requires a LOT of historical click data (query + answer pairs). Because Google has more historical data than anyone it has better results. I don't see how you can have better ranking without the same dataset to train a new model. Indexing the web is not the hard part.

Reply on news.ycombinator.com