Number of sites per lens needs to be increased.

MalcolmCook

5 seems too limiting to be of practical use.

Vlad

MalcolmCook We plan to increase the limit to 10 for paid subscribers. This should cover most niche use cases. If that still sounds like not enough what use case do you have in mind?

MalcolmCook

Vlad hi again - even without API & collab lenses I think I've made a case for more than 5 or 10. Do you agree?

I agree HN/Reddit would make a good lens. I would use that.

My Bioinformatics analysis lens would include reference manual for scores of libraries and applications, probably on order of a hundred.

But, this is speculative. As I said, finding the personalized results blocked domains don't currently work, I have refrained from testing other aspect of custom filtering/blocking/boosting. I'll ask again: do lenses work now? I look forward to putting them to use.

Vlad We plan to increase the limit to 10 for paid subscribers

Would you consider upping the count to 10 for beta users such as myself, at least till your beta period is over?

MalcolmCook

Well, since personalized results blocked domains is not working, I've not actually tried making lenses work. Do they work now?

Some of my work is performing bioinformatics analysis on genomics data as a collaborating member of a ongoing scientific project which may last years.

Often I am engaged in a handful of different such projects.

I could well imagine having a lens for each project.

Indeed I could well imagine wanting the ability to have a project lens to which all project members might collaboratively contribute.

Further, I could imagine wanting to have a lens being pdfs of all papers I have read as background to the project.

Actually, I am using SciWheel collaboratively on such projects. Project members add references to interesting project related papers to a shared cloud based repo, allowing mutual annotation (think yellow highlighter with notes in the margin) to occur in the cloud. There is an API to SciWheel from which URLs of discussed papers might be pulled to serve as a dynamic ever-updating lens as the project goes forward.

Of course there would be issues with IP & copyright, but you get the idea.

I'm not sure if this is pie-in-the-sky thinking. But you like such thinking. I find not many people want to collaborate in such fashion, but some do, and that number is only growing.

So add "Collaborative Lenses" to my hopes.

And an API to lens management.

And, and...

MalcolmCook

By the same token, what niche use cases do you image for wanting only 5 or 10 sites in a lens?

Vlad

MalcolmCook Almost all use cases I personally have are satisfied with a 5 site lens. For example: programming, recipes, local news or a HN/Reddit lens.

Collaborative lenses and API for lens management sound like two new issues you want to submit and detail.

Vlad

MalcolmCook There are compromises in execution where the complexity drastically increases beyond 10 sites in a lens.

What you describe is something entirely different. Lens is meant to search a whole site, even one that has million pages. 10 sites like that could be a billion pages searched. It is very different than indexing single documents for example for which you may need million entries in your lens.

MalcolmCook

Thanks. Obviously I am not making requests informed of the specifics of the engineering challenges in meeting them. Good to know this limit of 10. I will accordingly temper my expectations.

Also, not to nag, but:

I'll ask again: do lenses work now?

Would you consider upping the count to 10 for beta users such as myself, at least till your beta period is over?

Vlad

MalcolmCook Yes lenses work and we will soon increase the limit to 10 sites.

buteo

MalcolmCook I like to think of lenses as a powerful filter for surfaced results — they exaggerate some features of the original "signal" from the search indexes while attenuating other features. In my testing, several small lenses targetting particular aspects of my research question or even considered subsets of my literature resources were very effective. And by selectively enabling small lenses I also better understand the context for surfaced results. Oddly, the lenses most useful to me are those I created somewhat spontaneously after discerning simple patterns in my searches, e.g., "Not one result I've seen with keyword x offers anything of interest!" or "I always want to see site show up in the first page or two of results". I no longer try to overthink them.

NoGoogle

I have like 90 news websites that I rely on. Most are small non-mainstream ones that search engines don't pick up on. The current 5 or 10 sites in a lens limits my ability to search for news/information within only those sites. 90 sites in a lens would probably kill the search engine but if you were looking for a use case of more than 10 sites this would be it. 90 is the overall amount of news sites I use but if you were to break it down and only list the ones that do hardcore investigative journalism, I could easily list 10-15 of them. As a compromise, I could probably split them up into multiple lenses and do multiple searches assuming there's no limit on the amount of non-active lenses one can create.

jjrvi

@Vlad is there a technical reason for choosing the 10-site limit?

It seems to me that for things like recipes (if people want to make their own customized recipe lens with the sites they like) or coding or news, people will bump into the 10-site limit pretty quickly.