I guess it depends on your threat model, but there are also privacy concerns with a search engine you have to make a paid account for.
For a regular user who signs into Google to use Gmail, they'll be searching while signed in anyway - no real loss of privacy, and arguably much gained by not broadcasting anything into the adtech data pools.
For a very privacy conscious user who wants to be anonymous and is already taking steps like using VPNs, there is a loss in privacy due to payment info being attached. (You could argue Google accounts could be pseudo anonymous, but for most users they're anything but - location data, history, emails and search queries make them totally identified).
There are technical solutions to avoid the payment linking issue (cash in envelope, selling gift cards via retail, etc), or through using blind signature schemes to separate payment from usage. The user experience of these options is likely to be far more complex, with people losing access and customer service fiascos with people being unable to prove they signed up before etc.
Which is an option with Brave and DDG, but obviously not Kagi. Yes, obviously, this is only going to be a relatively fringe use-case that depends on the user having generally decent opsec.
No need to argue the abstract; all that matters is the specific. If Kagi specifically is abusing trust, I’d love to hear it.
If Kagi is being honest, typical usage of Kagi is probably more private than typical usage of advertising-supported search engines. But I can more easily verify that using e.g. Brave Search via Tor is private.
True. But even the payment thing could be solved although most users won't care. Mullvad comes to mind: you can send them cash via mail and the payment is associated to a serial number that is simultaneously your login and password.
That's not really the point. It is up to Kagi to prove to their customers that they are trustworthy, not up to the customers to find abuses of trust.
If I -- for example -- use Tor and a private browsing window to access Google search, I have a pretty good expectation of privacy. I don't have to trust Google here; they simply cannot correlate my search with my Google account in this case, at least not 100% reliably (browser fingerprinting, etc.).
But with Kagi, they know that every search I make, regardless of how I make it, is from me (or, at least, from someone using my account), and that account is tied to my real-world identity. Sure, they could implement a more complicated payment method, where I pay a third party, and that third party blind-signs some credentials that Kagi trusts. But they haven't done that, at least as an option, and that bothers me.
When it comes to third parties, the best kind of trust is the trust that you don't have to have. I have to trust that Kagi isn't keeping a record of my search results. I would rather that I just don't have to trust anything, because there's nothing nefarious they can do.
Fair enough. Since that is a provably impossible bar, you’re basically saying nobody should do business with anyone. Which I admire from an ideological purity perspective, but don’t really have time for at this age.
If I'm doing something in private browsing mode, I fall back to DDG. Not only do I feel weird about being signed into something when I'm supposedly doing something "private", either I have to manually sign into Kagi each time I open a private browsing window, or install their extension, neither of which I really want to do.
So I'm torn; I'm not sure if I'll keep paying for the account, or cancel it and go back to DDG.
Then again, I also pay Fastmail for email, which gives them access to all my email. Arguably that's just as bad (if not worse) for my privacy than giving someone my search history. It's not clear to me why I'm ok with that, but uncomfortable when it comes to Kagi. I suppose email is inherently tied to me, whereas search results need not be? I dunno.
Regardless, this is the big issue with paying for something on the internet that you can also get for free. In some ways my search privacy can be "trustably" better with Google (via private browsing + Tor or at least a VPN) than it can ever be with Kagi, which is weird.
Beyond the obvious privacy concerns I'm not going to bother repeating, it really depends how it is monetised - is it only used to target ads on other products or does the search engine itself have ads?
The former option is potentially fine as long as you're happy with the privacy concerns. The latter option (which is what Google is) directly incentivises them to degrade search quality so that you spend more time on the search engine and potentially click an ad.
Furthermore, when it comes to a large company such as Google that also happens to control ads/analytics on many other sites, it incentivises them to preserve ad/analytics-infested sites in search results (since they make money and/or collect data off it) despite those being very good signals a site is spammy and should be downranked.
(for antitrust reasons they will never explicitly talk about prioritising sites with their own ads or explicitly degrading search experience - however the unspoken rule is that projects/ideas that might improve the experience and lead to less ad impressions overall might not succeed for various innocent-looking reasons, giving them a layer of plausible deniability)
A user-funded search engine doesn't have those misaligned incentives and can beat SEO by using common hallmarks of spammy content such as presence of ads, analytics, affiliate links and other marketing tools (used to measure the effectiveness of the spam) as negative ranking signals. This will be difficult to evade by the spam sites because the same techniques they'd need to evade search engine detection (such as self-hosting and proxying the ads) are common in ad-fraud and would be prohibited by their advertisers, thus cornering the spammers in a game where the only winning move is not to play.