Hacker News Reader

Comment by jsheard | Hacker News Reader

Is there a reason why Google doesn't allow you to block specific domains from your personal results like Kagi does? Is it a performance thing?

Simply being able to block low quality domains from showing up again would go a long way to cleaning things up.

View on HN

Google: We need all your personal data so we can give you a better experience.

Me: Ok, please block Pinterest, W3schools, ...

Google: Sorry, can't do that.

What's wrong with w3schools ? They have come a long way since the w3fools days and their content is really good in my opinion.

A lot of the pages just scratch the surface on the subject to get ad revenue. The example will be incomplete, highly situational, only explain one out of five overloads, etc. Whereas something like MDN or an independent CSS blogger will list every argument, return types, every overload, have complete examples, etc.

They position themselves as a reference but are more interested in SEO rank and ad clicks than providing quality reference information.

To be fair it is probably a feature needed and wanted by 0.00001% Google users(compared to 90% Kagi users) so it gets prioritized away in PM meetings.

Also uniformity of experience is needed for large scale (trillions of datapoints) A/B optimizations of ad clickthrough rates. So less variabiity is better and this is why Google results look the same for everyone in the world.

In contrast, in Kagi world, user is in the center of it (as they are the ones paying for the product) and the vector Kagi optimizes for is user satisfaction. So users can customize not just what is in their results but also how the results are presented.

The extension uBlacklist allows you to do that (https://iorate.github.io/ublacklist/docs).

I think it may affect their ad revenue adversely, as users will just block website from which they see too many ads, in some cases people may just began to use it like a in-build ad blocker.

Even without custom domain blocking enabled, Kagi results are still somehow much much better (at least for my use cases). So clearly it’s not required to fix the seo spam problem.

My hot take on that is it's more profitable not to.

There was a time that "I'm feeling lucky" worked well because Google would often answer your query with the first result. But now, you often have to click through a few results, tweak your query, etc to answer your question. This almost always means more ad impressions delivered per query.

Delivering better search would mean delivering less ads, and that's the metric they value more.

I very much doubt it's a performance thing, because they already do personalized search based on account, location, etc, so it's not like it'd add any meaningful overhead.

There are lots of domains (like Wikia) I want to block 95% of the time, but don't want to totally block, in case I need them as a last resort.

Also many of the worst offenders seem to be spam sites with unique URLs. Too many for me to individually make a dent.

Kagi may interest you then:

https://help.kagi.com/kagi/features/website-info-personalize...

For a given domain, you can select: block, lower, normal, higher, and pin to help you rank results on your personal preferences.

Actually, it does, but you need to create a custom search engine. Strangely enough, even without banning the seo spam sites, it still gives better results than my vanilla google

Because they tried that already and it was gamed to death.

Gamed how? I can see how feeding users blocking decisions back into the ranking algorithm would be abused, but they don't have to do that.

Feeding results back into the algorithm was indeed the reason it was abused. 'They don't have to do that' is correct though, I don't know the reason for that.

Reply on news.ycombinator.com