7 ms·

I'm glad they brought up Marginalia; it's one of my favorite engines. Teclis[0] and Wiby[1] are similar contenders in the non-commercial search space. [0]: ht

by Seirdy 5y ago

I'm glad they brought up Marginalia; it's one of my favorite engines. Teclis[0] and Wiby[1] are similar contenders in the non-commercial search space. [0]: http://teclis.com/ http://teclis.com/ [1]: https://wiby.me/ https://wiby.me/ What the article mentioned about the influence of PageRank definitely rings true. An interesting variation is used by the Secret Search Engine Lab's CashRank algorithm[2]. [2]: http://www.secretsearchenginelabs.com/tech/cashrank.php http://www.secretsearchenginelabs.com/tech/cashrank.php I listed a bunch of engines with their own indexes, FWIW: https://seirdy.one/2021/03/10/search-engines-with-own-indexes.html https://seirdy.one/2021/03/10/search-engines-with-own-indexe...

loudtieblahblah 5y ago
it'd be nice if these could be worked into SearX
zzo38computer 5y ago
It look like good, but perhaps should have some more options (that can be specified as a part of the search query text), and documenting these features. Some possibilities would include filters (by file format, domain name, scheme, etc), sort order, excluding, etc. There are two menus for options ("Popular Sites", "Blogocentric Eigenvector", "Both Algorithms", "Experimental", "Allow JS", "Deny JS", "Require JS"), but does not explain them very well. (For example, I might want the search engine to not execute any scripts in web pages to determine their text, but if the text works when scripts are disabled that it can still be included in the search results (even if the web page has scripts, as long as those web pages work correctly even when scripts are disabled).) Also, they have some documentation using Gemini format. I have a Gemini viewer in my computer, but it won't use it because of the "Content-disposition" response header.
cryptolake 5y ago
I don't know, why i wasn't aware of these search engine, i'm having a blast discovering personal website, this is amazing.
aghilmort 5y ago
list is great -- & knew you'd done your homework when saw Marginalia - also a fave run search startup, Breeze, so have a similar list and I was like damn :)
- Seirdy 5y ago
  My feedback: I think what you're doing with Breeze seems interesting, but the value-add of making it a commercial offering isn't clear; what does it offer that anyone else can't easily replicate with Google Custom Search? I'm not saying that there isn't a value add, I'm only saying that it isn't obvious as a user. Something has to be a scarce resource; my guess is that the resource here is "labor" in building the CSE parameters and finding the sites to add to collections. Perhaps the effort that went into this should be emphasized. Are there plans to move things server-side? Making client-side requests to Google has privacy implications.
  - aghilmort 5y ago
    tackling in parts / next replies
    - aghilmort 5y ago
      -- Google / privacy / alternative proxies -- tl;dr -- working on improving client-side privacy protection & there's couple of other options that may permit using Google with a slight bit of user config; premium version is all server-side since proxied via Bing, Gigablast &/or our index; there's an alt premium option that would basically proxy google via a cloud browser, say browserling or KASM or similar 1. The client-side will be set to no personalized ads -- just found out about a fairly hidden setting that permits that -- should be changed / live later today 2. Google's API precludes making server-side unless user configs an account which we can then drop in, since limited to 10K queries / day to do anything meaningful custom or full web; we're planning to do that as blog post as intermediate alt 3. We've started to instrument if client-side calls sidestep any privacy -- those results / protections are necessarily limited, however, we can perhaps uncover some things if anything in their client-side code that is privacy revealing & possibly mitigate some 4. premium is server-side with proxy to {ahrefs*, Bing, Gigablast, etc.} OR our Breeze index -- we scrape inventory-sensitive / time-sensitive sites, e.g., used car dealer pages 5. given enough ad or other revenue, we could alternatively give everyone a Bing proxy like DDG or other services -- bit too bootstrapped to do that out of the gate, plus Bing has more constraints on custom search, so that's a mixed bag of outcomes 6. another approach, also necessarily premium, would be to proxy through a service like browserling for Google searches, since the API doesn't permit it at scale 7. premium also includes alerts, especially for things like say car dealer pages that are more time sensitive and harder to config than what free google alerts do
derbOac 5y ago
So I tried these search engines. They're fun to try but kinda remind me of search before Google. I'm frustrated with Google search results lately, they don't reach far enough back and are diluted with all sorts of crud. But these alternative search engines kinda just remind me of why people started using Google. When I use them, I don't get spam, but I don't really get things I'm interested in either. It's like they don't understand what I'm really interested in and I either get no hits, or get hits on things that are just completely unrelated. It's like old-school chat software or something, complete with all the computer misunderstandings and whatnot. Don't get me wrong, I'd love to see a lot of competition in search. I use DDG a lot. But I'm surprised at how positive people's comments are about these alt search engines, because to me they have problems on the opposite end of the spectrum. To me, what Google returns is a pretty good understanding of what I want, but corrupted by manipulative spam; the other ones generally return no spam, but also not at all what I want.
- ufo 5y ago
  With marginalia I think that is by design. They intentionally do not index any websites that are deemed "too modern", which includes most things one might be searching for. It does seem to be intended more for exploring unusual places.
- amelius 5y ago
  I just hope we can get a trust-based network combined with search. So if I trust some friends and search "best toaster", and one of my trusted friends has given a very high review score to a toaster, then I get that one as the top search result. Extend this to online communities, and you can ask "what laptop would HN recommend for Linux?", etc. Of course, there's a privacy issue to solve, but the functionality could be very useful compared to the crappy Google search results for commercial products.
  - aabbcc1241 5y ago
    I'm having similar idea but with less user involve: a browser extension that extract keywords from the websites you actually visited, and form a p2p database with your followers. If you see spam / undesired ads in search result, you can rate down the content, and the system auto reduce the weighing from the peer providing that history. To avoid sharing sensitive page / habit, maybe let the user review in batch and confirm before sharing out the list.
marginalia_nu 5y ago
Honestly kinda crazy the publicity my tiny project is getting for what it is.
- Seirdy 5y ago
  You deserve every bit of it, and then some. The scope of Marginalia may be small, but it is one of the best examples of making the non-commercial web accessible to a much larger audience. I hope it inspires others to tackle similar projects. I'll be able to support the project starting a few weeks from now, and would love some non-Patreon options.
  - marginalia_nu 5y ago
    What is a good alternative to Patreon? I haven't really looked into it too deeply. They do take a fairly steep cut, and that's not even considering that PayPal also wants their pound of flesh on top of that :-/
    - Seirdy 5y ago
      Ko-fi and Liberapay are platforms that don't take fees. Buy me a Coffee is a popular choice for one-off donations.
    - leobg 5y ago
      Why not set up your own Stripe?

by Seirdy 5y ago

loudtieblahblah 5y ago

it'd be nice if these could be worked into SearX

zzo38computer 5y ago

It look like good, but perhaps should have some more options (that can be specified as a part of the search query text), and documenting these features. Some possibilities would include filters (by file format, domain name, scheme, etc), sort order, excluding, etc. There are two menus for options ("Popular Sites", "Blogocentric Eigenvector", "Both Algorithms", "Experimental", "Allow JS", "Deny JS", "Require JS"), but does not explain them very well. (For example, I might want the search engine to not execute any scripts in web pages to determine their text, but if the text works when scripts are disabled that it can still be included in the search results (even if the web page has scripts, as long as those web pages work correctly even when scripts are disabled).) Also, they have some documentation using Gemini format. I have a Gemini viewer in my computer, but it won't use it because of the "Content-disposition" response header.

cryptolake 5y ago

I don't know, why i wasn't aware of these search engine, i'm having a blast discovering personal website, this is amazing.

aghilmort 5y ago

list is great -- & knew you'd done your homework when saw Marginalia - also a fave run search startup, Breeze, so have a similar list and I was like damn :)

Seirdy 5y ago
My feedback: I think what you're doing with Breeze seems interesting, but the value-add of making it a commercial offering isn't clear; what does it offer that anyone else can't easily replicate with Google Custom Search? I'm not saying that there isn't a value add, I'm only saying that it isn't obvious as a user. Something has to be a scarce resource; my guess is that the resource here is "labor" in building the CSE parameters and finding the sites to add to collections. Perhaps the effort that went into this should be emphasized. Are there plans to move things server-side? Making client-side requests to Google has privacy implications.
- aghilmort 5y ago
  tackling in parts / next replies
  - aghilmort 5y ago
    -- Google / privacy / alternative proxies -- tl;dr -- working on improving client-side privacy protection & there's couple of other options that may permit using Google with a slight bit of user config; premium version is all server-side since proxied via Bing, Gigablast &/or our index; there's an alt premium option that would basically proxy google via a cloud browser, say browserling or KASM or similar 1. The client-side will be set to no personalized ads -- just found out about a fairly hidden setting that permits that -- should be changed / live later today 2. Google's API precludes making server-side unless user configs an account which we can then drop in, since limited to 10K queries / day to do anything meaningful custom or full web; we're planning to do that as blog post as intermediate alt 3. We've started to instrument if client-side calls sidestep any privacy -- those results / protections are necessarily limited, however, we can perhaps uncover some things if anything in their client-side code that is privacy revealing & possibly mitigate some 4. premium is server-side with proxy to {ahrefs*, Bing, Gigablast, etc.} OR our Breeze index -- we scrape inventory-sensitive / time-sensitive sites, e.g., used car dealer pages 5. given enough ad or other revenue, we could alternatively give everyone a Bing proxy like DDG or other services -- bit too bootstrapped to do that out of the gate, plus Bing has more constraints on custom search, so that's a mixed bag of outcomes 6. another approach, also necessarily premium, would be to proxy through a service like browserling for Google searches, since the API doesn't permit it at scale 7. premium also includes alerts, especially for things like say car dealer pages that are more time sensitive and harder to config than what free google alerts do

derbOac 5y ago

So I tried these search engines. They're fun to try but kinda remind me of search before Google. I'm frustrated with Google search results lately, they don't reach far enough back and are diluted with all sorts of crud. But these alternative search engines kinda just remind me of why people started using Google. When I use them, I don't get spam, but I don't really get things I'm interested in either. It's like they don't understand what I'm really interested in and I either get no hits, or get hits on things that are just completely unrelated. It's like old-school chat software or something, complete with all the computer misunderstandings and whatnot. Don't get me wrong, I'd love to see a lot of competition in search. I use DDG a lot. But I'm surprised at how positive people's comments are about these alt search engines, because to me they have problems on the opposite end of the spectrum. To me, what Google returns is a pretty good understanding of what I want, but corrupted by manipulative spam; the other ones generally return no spam, but also not at all what I want.

ufo 5y ago
With marginalia I think that is by design. They intentionally do not index any websites that are deemed "too modern", which includes most things one might be searching for. It does seem to be intended more for exploring unusual places.
amelius 5y ago
I just hope we can get a trust-based network combined with search. So if I trust some friends and search "best toaster", and one of my trusted friends has given a very high review score to a toaster, then I get that one as the top search result. Extend this to online communities, and you can ask "what laptop would HN recommend for Linux?", etc. Of course, there's a privacy issue to solve, but the functionality could be very useful compared to the crappy Google search results for commercial products.
- aabbcc1241 5y ago
  I'm having similar idea but with less user involve: a browser extension that extract keywords from the websites you actually visited, and form a p2p database with your followers. If you see spam / undesired ads in search result, you can rate down the content, and the system auto reduce the weighing from the peer providing that history. To avoid sharing sensitive page / habit, maybe let the user review in batch and confirm before sharing out the list.

marginalia_nu 5y ago

Honestly kinda crazy the publicity my tiny project is getting for what it is.

Seirdy 5y ago
You deserve every bit of it, and then some. The scope of Marginalia may be small, but it is one of the best examples of making the non-commercial web accessible to a much larger audience. I hope it inspires others to tackle similar projects. I'll be able to support the project starting a few weeks from now, and would love some non-Patreon options.
- marginalia_nu 5y ago
  What is a good alternative to Patreon? I haven't really looked into it too deeply. They do take a fairly steep cut, and that's not even considering that PayPal also wants their pound of flesh on top of that :-/
  - Seirdy 5y ago
    Ko-fi and Liberapay are platforms that don't take fees. Buy me a Coffee is a popular choice for one-off donations.
  - leobg 5y ago
    Why not set up your own Stripe?