List of websites mistakenly indicated as paywalled

carl

As suggested by Vlad in another thread, I will list websites I encounter that are mistakenly marked as paywalled. If anybody visits these sites and are presented with a paywall, please write in the thread.

I begin with:
~~https://madame.lefigaro.fr~~
(It had at least one paywalled article)

KagiForMe

But what about this article: https://madame.lefigaro.fr/bien-etre/psycho/comment-positiver-en-cette-epoque-anxiogene-la-lecon-d-espoir-de-la-philosophe-laurence-devillairs-20231210

It is clearly paywalled.

carl

KagiForMe Good catch! I tried at least 8 articles on the domain, none being paywalled. So, is the mark supposed to be there when there is the slightest risk of an article being behind a paywall, and how useful will it be in that case?

The intention of the thread is only to list domains that do not have paywalled articles, but are marked as such.

KagiForMe

This is what I meant in the other thread: All of those sites contain at least some paywalled articles. It would be wrong to just mark it as non-paywalled. The tooltip says "might be behind a paywall". I am against the entire removal of this tag.

I think there is some confusion regarding the scope of the feature "Show paywalled articles". The list that is used is not listing individual articles. There is probably no list at all for that (as the task would be too large). Instead, it lists websites which have at least some paywalled content. This is all the feature (i.e. symbol) is showing. Everything else is not in the scope.

I thought that this was clear from the way the discussion was handled in the proposal..

To quote Vlad:

I think we are agreeing that it is not possible to say if the article will be definetely paywalled/login walled, but just indicating such domain would be useful. The easiest way to do this is to use a list of known paywalled/login domains[...]

Maybe @Vlad can clarify (confirm? deny?) this as the intended scope?

(@carl I don't want to be offensive, I'm merely disagreeing with you 🙂)

KagiForMe

Oh, and by the way:

If there are indeed false positives (or negatives) on the list, than I believe that this should be fixed upstream (i.e. filed against this repo). This seems in line with Kagi's mission and values and everyone profits.

KagiForMe

Ah, yes, that is justified criticism. I think it is the best one can do. It still is useful to see it at a glance which ones might be paywalled and whoch ones aren't. Personally, I use this as a quick indicator whem I have to decide on which article I click, supposed they are both promising.

I don't think there are any websites on the list without paywalls, aa the list is highly maintained (last updated 8 hours ago).

carl

KagiForMe Yes, then they seem to be on top of things, and no need for this thread. However I don't know exactly how useful this feature will be for the search results, unless there is some more fine-grained control than to mark every domain that has at least one paywalled article. And as far as I know, the repo only lists domains that they've successfully circumvented, is that correct?

Could Kagi have its own list, with other conditions?

KagiForMe

carl

And as far as I know, the repo only lists domains that they've successfully circumvented, is that correct?

Well, yes and no. Most of them are not circumvented, but list a link to some internet archive that has the full text for search engines. So this list is actually a general list, not a specific "we can circumvent this with tricks".

Could Kagi have its own list, with other conditions?

I am not so sure about that. From Kagi's perspective, this seems like a lot of development costs. You have to maintain this list, make a repo for it, accept pull requests, etc. What percentage of paywalled articles is enough? 30%? 50%? What if only specific topic (politics? sports?) are paywalled? It just seems like an impossible tasks to do "right".

Vlad

Correct, this is why it is a 'best effort' thing on our end. Happy to take further guidance from community.

KagiForMe

@Vlad (I think this can be closed, as the issue turned out to be none. If we want to keep a discussion, we should open a new issue, as the issue and title are not really a general discussion thread.)