Steps to reproduce:
Search for something like reddit.com technoproduction
Expected behavior:
Get more than one page of results
Debug info:
Windows, Firefox
Image/Video:
29
Add multiple pages of results
Something like this should return a lot more results or at least prompt the user to request more. There are hundreds if not thousands of posts. When I try to be more specific it's even worse.
Merged 2 posts from Small result set.
Steps to reproduce:
Run this search:
https://kagi.com/search?q=uk+5ghz+wifi
Only approximately 20 results are returned. Only a few of those are specific to the UK, and an appreciable number are related to shopping, not a discussion of 5GHz WiFi channels in the UK.
Expected behavior:
The same Google search returns 3.5 million results! The first page of results, at least, are specific to the UK and none are shopping-related.
Granted, probably many of those Google results are spam, but I would expect there would at least be a few thousand informative articles and sites that cover this topic.
Debug info:
I've tried this on Safari and Chrome with the same results. I thought it must be a content blocker hiding the "next page" button, but that does not appear to be the case as I've reloaded the page with content blockers disabled.
Image/Video:
dhess Our goal is to help users find what they are looking for not show millions of irrelevant or spam results. With search, quality>quantity.
https://kagi.com/faq#relevance
Did you find the result you were looking for this query?
3 months later
Merged 2 posts from Only 17 results returned for a search that should generate orders of magnitude more hits.
- Edited
I think that philosophy is making you blind to the use case where a user wants to browse dozens if not hundeds of links as a way of discovering more content. If you give a limited window into a search query you are assuming the user only wants that many results. However when researching things, it is quite common that the amount of things that are of interest exceed a dozen or two dozen pages. If you strictly limit the result set without being able to lengthen it with an additional request I will have to say that this is a deal breaker for me. I very often require more than a page of results in my research.
Also, the amount of "personal" sites that discuss these items are rarely even indexed as much as they used to be in the younger days of search; and I am certain they have valuable content and are in the hundreds if not thousands. When I am not just researching or trying to learn, those additional vectors are very important. We don't want to encourage echo chambers do we?
Please maintain your bar of excellence but also grant the user who wishes to view more, to view more. If need be even say that the results after the initial set are of less certain quality. It's frustrating from a user experience to be in a position of helplessness when you know where is more and no action can be taken.
I support the vision and high quality you are striving for - I'd just like to be able to handle the use case of discovery and exploration as well. There is a key consideration of echo chambers and recommendation engines forcing a local maxima.
aaaaddddd Your point is well taken. One thing we need to consider at Kagi is cost of search.
While in an ad-supported search engine more pages = more money (more likely to click on ad) with Kagi more pages = more cost.
If we added the option for more results and people searched 10 pages vs 1 per query our cost could explode. This is why we need to balance until we have pay per search model or a higher base price.
I know it is not the answer you wanted but it is something we need to think about.
a month later
@Vlad I can imagine a scenario where paginated results actually reduce costs: instead of displaying 20–50 results by default on one page, display 10 results per page; most users may be satisfied with results on the first page. Wouldn't 10 results cost less than 20–50?
Yes, some users would continue through 5+ pages, but presumably the cost for additional pages would not be more than requiring the user to submit a new, rewritten query that then loads another 20–50 results.
- Edited
Wouldn't 10 results cost less than 20–50?
Not exactly, our current result count comes from the combined results of distinct API calls & lookups in our own indexes. It is this mixing of indexes that give Kagi its quality. We are charged per-API call, so we request the maximum. We are charged the same if we request less results from each index.
The only savings is bandwidth, which is why we offer the "result count" option in your settings, so we send you less data at once for i.e. saving mobile data usage / slower connections.
a month later
Vlad I encountered this issue today. Looking for information on something very specific (orthopaedic thumb splints) Kagi's 25-30 results just weren't enough. I had to swap back to Google for what I needed.
It's not because I was looking for one good result that was buried three pages deep. What I needed was to find at least 20 different pages, of different manufacturers and retailers. Most of them weren't on Kagi's single results page.
If your use case is "the user wants to find the best link, Kagi will have it on the first page", then the Kagi model works. But once in a while we need to do a search where we want to look at dozens of results, for legitimate reasons, not just browsing because we are bored or distracted.
In fact I probably cost Kagi more because I did about half a dozen fresh searches with tweaked keywords before I gave up and went to Google.
With respect to the cost issue, if you use a design pattern to discourage users from casually clicking more, I wonder how many extra requests you would actually see... is it possible it would be insignificant?
We added More results with 30-50 additional results!
No one is typing