Show community rating of website in "shield popup"

iynaejVcJ

Kagi has a great source of "helpfulness" by looking at the "Domain insights" (as shown in the live stats). It would be great, if an aggregated domain rating based on user popularity could be shown in this popup:

That way I could better understand if other users deem the site trustworthy or not.

The rating could be calculated based on the blocked-to-pinned ration, taking sample size (and variance) into account.

The rating could be shown like in this quick mockup:

Maybe, with an additional tooltip, showing what it is.

Or just as a shade, representing frequency of usage of block level (some trickery would be required to get a good value for "Normal" as that is the default and would probably always be black, without adjusting the value somehow).

The shown screenshot shows a website with a website that the community rates as slightly less relevant.

carl

May I suggest that Kagi waits with such a feature until the user base is a bit bigger and broader? Looking at the blocked and raised domains on the stats page, I get the impression that many users are not ranking domains on usefulness, but on political zeal. Maybe even going down the list and blocking websites they hate for political and ideological reasons, without the domains ever having showed up in their results.

For example Instagram is one of the top 10 blocked domains right now. I have almost never had a non-useful Instagram result in my search results, on the contrary many times the best result is on Instagram, when searching for a restaurant or concert for example. So I'd suggest to wait until Kagi has at least 100 000 users before starting to integrate block/pinned stats.

Anonymous12

I think it's a very good idea. Having the data more easily at hand is precisely what those domain insights are for.

Vlad

iynaejVcJ Interesting idea, I like that it could surface Kagi stats better. One challange I see there are domains that are both blocked/lowered and raised/pinned (usually political stuff and reddit).

I guess what I am asking is what would the algorithm be?

The second issue with this is only a handful best known domains would be ranked as Carl says, which arguably people never "info-hover" over and those that they would, would have no such signal..

lix44

This information is already accessible to users, but you're enhancing its availability. Community ratings may never be unbiased. For instance, since most tech enthusiasts (who form the majority of Kagi's users) dislike Meta, Instagram may consistently receive low ratings. However, you can disregard these ratings if you're specifically looking for Instagram results. This tool is most valuable when exploring unfamiliar results; you can then investigate why something is poorly rated.

WoundedMoose

Because one finds what is true based upon popularity? Utter nonsense.

I'm all for the availability of "data for nerds", but pretending this is anything other than the attempt du jour to get grading/filtering/censorship by something other than a source's actual merit is dishonest.

taperropein

Hi, new (paying) Kagi user here.

@Vlad , you seem to be worried about the ecological impact of Kagi, given your comment about renewable energy in the other thread, so you definitely realize that externalities are something to worry about.

I hope you (and everyone who thinks that "usefulness" is orthogonal to politics) will read this account of the human cost of Facebook's entry into Myanmar, because as someone responsible for providing access to information, you have the power to cause something just as catastrophic... or prevent it from happening.
https://erinkissane.com/meta-in-myanmar-part-iv-only-connect/

Most people in your position try to deflect this responsibility, either because it's in their short term interest (long term social interests be damned), or because they are simply intimidated by it.

I hope you will not fall into the same trap as them, I would like to believe that the company I'm supporting financially is not as spineless as all the megacorps that would rather destroy society than refuse to do business with anyone not on a sanctions list.

taperropein

lix44 Look up "instagram sex trafficking" if you want to have a bad time.

Maybe Instagram is useful to you, but would you be comfortable directing a 12 year old there?

WoundedMoose

taperropein Kagi is affiliated with Facebook how?

lix44

WoundedMoose it is not

WoundedMoose

lix44 Correct. Thus @taperropein is muddying the waters and can be ignored.

taperropein

WoundedMoose Maybe I'm not using the reply function correctly, which is why it's not immediately obvious what I was responding to, but @carl was talking about the usefulness of search results that direct you to Instagram.
My issue is that @carl might only see the nice side of Instagram and not be in danger of being groomed by sex traffickers, but for others (mostly young girls) it would be very useful to know that a site in their search results hosts sex traffickers.
A search engine's whole purpose is to help people find relevant information, but you won't know what counts as "relevant" unless you think about other people's perspectives. I presented a perspective that most tech enthusiasts (who currently make up the majority of Kagi's user base) don't know about, because it doesn't affect them.
If you consider that to be muddying the water, then I don't think you are engaging in this conversation in good faith.

Vlad

taperropein Every sufficiently large social media site in the world probably has some users that are sex traffickers. As it does host crimianls etc. That should probably be common sense not sure that we need to specifically highlight this? Or I am missing your point.

I like the original idea just trying to find ways to make it actually useful due to constraints highlighted here https://kagifeedback.org/d/2845-show-community-rating-of-website-in-shield-popup/4

lix44

Vlad I don't think it should affect the algorithm at all, it should only have an informative purpose.

Value7609

Vlad doesn't the second suggested mockup from @iynaejVcJ combat this challenge?
By highlighting the bars below Block/Lower/Normal/Raise/Pin in a color between white (0%) -> black (100%), then if something gets both lowered and raised, it is still visible that both options get used a lot, compared to an average which in this case may be "Normal".

Otherwise, it could also be just text, e.g. if something gets lowered/blocked a lot: "20% of kagi users don't like this page" or if something gets both lowered and raise: "kagi users can't decide if they want this page in their results or not". I'm not good at writing English sentences, even less in corporate speech, but I hope these examples were enough to understand my idea.

This feature definitely needs a little info button that explains how this "rating" is determined.

iynaejVcJ

WoundedMoose I don't see how my suggestion is censorship. Also, who says, "true based upon popularity?". I only say, I would find other users blocking behavior useful to know when dealing with websites I am not familiar with.

Maybe there is a misunderstanding here: I am not asking for the search ranking to be changed. My suggestion is towards displaying data in the shield pop-up. Data that already exists, but is currently less accessible.

Vlad said:

One challange I see there are domains that are both blocked/lowered and raised/pinned (usually political stuff and reddit).

I guess what I am asking is what would the algorithm be?

I think, incorporating some kind of confidence (e.g. the variance) would help. Regarding my first mock up, we could compute the average user rating and divide it by the variance (and some constant). That way, results that are blocked by many and pinned by many would show a less extreme (closer to 0) community rating (assuming 0 is normal). (The variance would have to be clamped to values between e.g. 1 and 5 or so)

Or, using the second mockup, just showing the absolute values, the user can see that a domain is controversial themselves. We could even display the absolute numbers of "blocked" etc. (with colored background, so it can be understood at a glance).

WoundedMoose

iynaejVcJ I did not say you are attempting censorship with this, I am lumping you in with others who want, once again, some sort of filtering/ranking/censorship, so Site I Don't Like can be written off. Every week there's some variant of this, because unbiased search is a threat to people who begin with their own conclusions.

And yes, you are saying a website is more true based upon popularity. It's exactly what you're saying. You want the crowd to think for you.

WoundedMoose

taperropein Other people's perspectives have no impact on what is true or what I find useful.

taperropein

WoundedMoose Saying the quiet part out loud, huh.