Reputation/reliability of site in site details

LasathFernando

What does your feature entail? What is it for? How will it affect existing workflows or user experience?
It's great that I can hover over the shield icon to get a summary of a website on the results page.
However, I wonder if it's also possible to include that sites history of misinformation? Whether it's a reputable source for something? If its the authoritative source for a given topic?

I understand that it's not an easy feature, but it would really set kagi apart from everyone else in this era of information warfare.

What are the exact ways that you see a user using your proposed feature? Please go into as much detail as possible, and provide examples of how other browsers/apps implement this feature, if applicable. If your feature suggestion adds on to an existing feature, how would it work into it to extend its usefulness?
It would show up here, along with the rest of the site details:

Vlad

LasathFernando How would we determine this?

LasathFernando

Depends. Is the "quick answer" powered by a knowledge graph (like Google), or a large language model (like Bard/ChatGPT)?

If it's a knowledge graph, I kind of trust it to give reliable information:

From there, maybe you could build a meta knowledge graph that tracks which sources are linked to / have a controlling interest in what. Then use that to build ranks / information on the sites.

If it's actually powered by an LLM, you might need to start by building a knowledge graph instead 😛 (I'm not super familiar with the latest developments on AI/ML so I can't suggest a concrete plan if that's the case).

Since you're indexing everything anyway, it would also be useful to collate/compare articles and give a score for how much a given one deviates from the consensus. That alone would give me an idea of which one(s) I want to read from a list of results. You can also compare to articles with known biases as extra signals.

Anyway, I'm sure the actual experts at your company know better than a rando on the internet. I'll quit my rambling for now. 😃

Vlad

LasathFernando Good ideas. We do not index everything though, and would prefer to leave the rating of websites to our users (which they can through personalised results) .You know the old saying "one man's trash is another man's treasure" and it feels that this is something that is best not done by corporation.

searchboy

carl while the absolute truth is never guaranteed, it

Vlad one man's trash may indeed by other man's treasure, however most people share the same trash/treasure mappings, that's what make treasures expensive🙂
Maybe crearting another 'lense' that take site's popularity/credibilty into account may work?

LasathFernando

I see. I didn't know Kagi was a meta-indexer at the time.

But that being said, in this era of rampant "Fake news" and other information warfare, it's important to remember that there is in fact, an objective truth.

If Kagi can work on helping people determine which sources are scientific, unbiased, authoritative, etc, that would be a tremendously useful tool. 🙂

carl

LasathFernando

For people who don't have the mental capacity to separate fact from fiction, what does it matter if they have the correct opinion or not?

Fake news have been with us as long as news have. Usually the fakest news have come from the most authoritative sources with the most consensus. That's how wars usually are started.

I also notice that during recent years, consensus has been made much more important than honest searching for the truth, meaning yesterdays fake news become today's truth and yesterday's truth becomes today's evil conspiracy theories and fake news.

I don't really want to see my Kagi results distorted in the way you're suggesting.

Vlad

LasathFernando

If Kagi can work on helping people determine which sources are scientific, unbiased, authoritative, etc, that would be a tremendously useful tool.

This is what we have been doing and we found that counting the number of ads/trackers is a very good indication fo the quality of content. In other words you are asking the question what was the purpose of the created content? If it was to drive page views/ads then it is likely to be of low infomrational/educational quality (most mainstream media is example of this).

And we already have this indicator in site info. It is very hard to establish any other signal that is both as powerful and as scalable to implement.

ntt

searchboy Popularity≠Credibility and popularity are often not even useful. Since there is no relationship between popularity and quality, tracking the number of clicks on sites and ranking them by popularity, just like Google does, gives terrible results, as we do now. Just look at the list of the most popular sites in the world and you will see how many sites are proven to be harmful to human psychology.

searchboy

ntt agreed, popularity≠credibility.
I guessed creating a credibility measure is much harder while popularity measure should be easier.