If you think through the feature, it's not at all obvious that it's a good idea for an established search engine targeting the general public [0]. But even then it might be a great idea for a search engine that's trying to establish an initial niche. It doesn't even matter whether the idea actually works well in practice: if 100k people think that manual curation of search result domains is a high value feature and are willing to pay for it, the feature has done its job.
[0] I wrote a more detailed comment on why a few months ago: https://news.ycombinator.com/item?id=34294800
Which suggests to me that while kagi might be good now, if it gets popular it may undergo enshittification in an effort to monetise its popularity.
For Kagi, it's a matter of when, not if. Capitalism guarantees it.
Both are technically possible, but so extremely unlikely, that you can rely on them not happening.
I concede that if they got big enough, someone would come with enough money that they couldn't refuse, and buy them to try and "unlock vale" by racing to the bottom.
Don't confuse an entire economic system with the unique problems caused by public ownership or investing (i.e. growth at all costs). Plenty of smaller organizations would be quite happy to operate with modest profits for their owners and never scale beyond that point.
Precisely, which is why enshittification is a when, not an if. A paid search engine will never pick up enough users to survive at $10/mo. It will either be acquired or start throwing in "non-intrusive" ads to balance the books.
Here is a personal guarantee that this will never happen with Kagi while I am in charge.
There are too many ad-supported search engines and browsers out there. I would not waste 10 years of my life to make yet another one. If Kagi can not sustain iteself with memberships, it will be the end of it.
The initial reason I started Kagi was to provide my kids with an opportunity to grow up in an ad-free experience of the web. In this interview [1] I talk more about my motivation.
Call us back in 6 months when you arbitrarily change something else and give us another motivating pep talk about your commitments. roll eyes.
I was just about to finish this post and ask if anyone knows the history. I Googled that and was surprised by the answer from Wikionary: https://en.wiktionary.org/wiki/enshittification
...[C]oined by Canadian-British blogger, journalist, and science fiction author Cory Doctorow in 2022.
Kagi has already put itself on a path where it doesn't have to make itself worse to cover its expenses, because it is already covering them.
Imagine if one of your friends come back from a holiday trip and tells you how fantastic the place they visited was. Is your response "Well, maybe that place ain't going to be so nice in a few years!"?
The usefulness of a per-user block list is not fighting outright spam. The statistical filters, etc, as applied by the search engine, work best here, as you correctly state.
The point is to remove certain well-known sites with information about well-known topics, which is unhelpful for the particular user, me. That information is technically not spam, but it's on a wrong level, badly presented, etc.
That is, if I search some info about a particular CSS trick, I want MDN be on top, some professional blogs be near top, and w3schools be on the third page. For someone else, priorities may be different, but I know what to expect for the search queries I issue, and I'd be glad if the search engine helped me.
Equally, certain domains are known to be more rich with information I happen to seek, like reddit.com, so I'd like to give them a boost. Some other people may prefer something else, or not give any boost at all.
So, this is personalization, not anti-spam. If Google kept it, it could even bring important insights into what ads I would skip and what would click :)
Maybe 90% of users would ignore this feature. But the 10% of power users maybe disproportionately important: with more disposable income, more clout on social networks, more decision-making power. I suppose that the shutdown of Google Reader, which affected a relatively small number of users, affected all the wrong people, and destroyed so much goodwill that the losses from it are many orders of magnitude larger than whatever savings from not supporting it any more.
I'd count that kind of low quality site as spam, and the problem is that there's a basically an infinite supply of them. The boost feature is more interesting in that the curation is at least not a never-ending task. New high quality sites that I know about don't pop up all that often!
It's still not an entirely robust feature since you'll need to do a lot of quality work to prevent mediocre results from a boosted site from shadowing better results from unboosted sites, especially for users with a ton of boosts. But that quality work is much easier to justify when the manual curation is one of your core selling points and used by a large proportion of the customers vs. when it's a feature in a dusty context menu that basically none of the userbase interacts with.
> Maybe 90% of users would ignore this feature. But the 10% of power users maybe disproportionately important
I'd be shocked if the number of users for this kind of feature was anywhere near 10% of all search engine users. Something like 0.1% seems much more likely. But again, if those 0.1% really want that feature it's a great fit for a power user search engine.
It still doesn't mean that it's a feature every search engine should obviously have.
That strikes me as a non-goal. As the search engine, you can't know what attributes a user is valuing when they boost, and so you can't know when it's appropriate to override that boost. "Better" is subjective, contextual, and comes in thousands of different flavors.
But when they do, how are you going to know if all you see is the mediocre spam instead? No, I want to block the domains which are never going to be a good result. This may actually surface things I will want to boost.
NYTimes is another good example, great content for subscribers, essentially useless spam in a search result for non-subscribers. The only person who can configure this is the user.
Is their information of such low quality ?
They’re better now but this is timeless.
I don't think he did. He did say " it's not at all obvious that it's a good idea for an established search engine targeting the general public"
And looking at the data from the link, other than wikipedia and reddit, all the pinned domains are for developers or technical folk.
So it looks to me that kagi is not popular to the general public. If it was, programming-related sites wouldn't so heavily top the popular list.
When you’re first learning a subject you often want simple lies more than complicated truths.
Like pinterest.
Goddamn pinterest.
Just the amount of times i've used "-pinterest" in search is staggering.
Now, I'm not blind to the fact that Reddit's user demographic has a lot in common with us HN folks, while Pinterest... well, not so much. It leans towards a non-techy crowd, with more women users. But the way I see it, if Google keeps popping Pinterest up in search results, someone out there must be finding it helpful.
Perhaps what this all boils down to is a bit of a demographic reality check, rather than a definitive statement on any one platform. And just to be clear, I'm not standing on a soapbox championing for Pinterest, but I reckon we could probably nudge Google to get a bit smarter with its algorithms.
So, what if instead of giving domains the chop, we pushed for Google to up its game on the personalization front? Like, if I'm consistently booting Pinterest from my searches with "-pinterest", Google could take the hint and drop it down the ranks for me. Wouldn't that make more sense than completely blacklisting sites that some folks might still want to use? In the end, we all win if Google gets better at tuning into our search habits, right?
Same for youtube, a simple blacklist would make my search results much nicer, but youtube seems to push a few large media companies as hard as possible.
> Third, some people will probably block domains they shouldn't have blocked, and then have a bad user experience in future searches as the sites with genuinely best results is blocked. And then you're only left with only bad options: ignoring the users' stated preferences which they'll hate, or serving bad results that they'll also hate.
That's a real possibility, but it can be addressed by presenting a block that would say, for example "results from blocked domains" or some such.
They primarily used "Experts Exchange" and http://www.experts-exchange.com/
This is bad for Google's bottom-line thus why such a feature was removed and will never be coming back.
1. Sell the first few results but make it look organic (this all happened behind closed doors and I knew a few people who paid for such results)
2. Hierarchical/topic based collections of links (This is what Yahoo was famous for in the 90s if I remember correctly, I'll keep saying that because it was a long time ago!)
The whole "google cares about you" was an outcropping of the organic flocking to google because of their reliable and algorithmic results that just seemed trustworthy, they capitalized on that and it honestly lasted a lot longer than I expected before they really started being evil.
Google didn't prioritize it's visibility [intentioally] over the past 2 decades as well. Maybe it's time to do so.
I can't imagine how tedious the process may be to make very small changes on the main google home page.
The features you were referring to were removed a few years after they were launched.