I think that Searx has a lot of potential but is rather bad if one uses the default values, so, I’d like to comment several issues with the hope that these “rants” may contribute to make Disroot’s search better. I’m aware that it’s probable that some of these rants aren’t fault of Disroot’s installation and configuration of Searx on their servers, but I don’t know which ones depend on the software’s configuration and which ones are product of Searx’s code exclusively. So, excuse if I comment about things that are far away from what a simple configuration of instance can do.
I hope you excuse the lenghth of this post.
I observe with stupor that almost every piece of free (libre) software, that software which should not only bring freedom to our technology, but also make said technology more accesible, democratic, neutral, respectful with persons, cultures and communities, against language barriers and against advantages for native speakers of this or that language, instead, keeps on supporting and reinforcing this, not innocent nor uninterested at all, model of an Anglo America-centric world (when a search engine plays favorites with a language, makes its users visit websites from that cultural -ergo most probably economic- zone and make those sites gain visitors, reputation, users, clients, etc; and all this, in detriment of other zones of the world. You think that if Amazon had been polish or Facebook japanese they had had the same success?).
Unfortunately, Searx doesn’t escape from this digital form of neocolonialism: no matter where you come from (IP address) nor in which language you have set your browser (browser’s user agent), the search results you get are in en-us. This neoimperialist behavior is so undissimulated that even if I search for terms in my language (not a single word in english), even about my country, my city (couldn’t be more local!), and even with the search terms in quotes, I get results in english in first place and in majority! o_O I have to change this behavior in the drop down menu or in Preferences to get the REAL “default language” for me.
This isn’t just a matter of “cyberactivism” in pro of human diversity and equalty. The search results are affected in their quality and utility due to this issue: like most persons, I only use english when I have no choice, in my everyday life I use my native tongue, and despite I’m able to slow and clumsily communicate in english, the language I handle well, rapid and comfortably, in which I think and speak is my mother tongue, thus, when I do a web search I do it in my language (sometimes I do it in english if I want to read about technology, but not because I want, but because there’s much more and deep technological publications in english, this forum is a proof of it). Searching in one’s native language not only makes the results comfortable and agile to read, but also limits the geographical area: if I search for a portuguese restaurant, most probably I don’t want a portuguese restaurant in Los Ángeles, Ohio or Jamaica but in my city; if I search for some local news I don’t want be informed by some foreign media which doesn’t have good first hand info and comprehension about my country nor can even spell correctly my language; if I search info for job opportunities, I most probably want them in my city or at least in my country, etc, etc. So, besides, the search results are bad.
Searx shoud do like other engines and use the info provided by the browser’s user agent locale, which most probably is not en-us for most of the humanity. Curiously enough, Searx seems to respect a bit my locale since it shows its interface in my language O_O.
So, since Searx detects that I don’t have american english as the language of my browser ergo I most probably am not looking for results in en-us but in my browser’s language, if this can be tweaked in some configuration script, please configure your Searx instance to respect each user browser’s locale sttings. If I punctually want results in english, chinese or swahili, dont worry, that I’ll choose them from the drop down menu by myself.
This leads me to suggest a software feature: an option to select a group of languages to search.
Most of us can read more than our native tongue and english, especially in the same family of languages, even if we can just understand half of the text and can’t speak a word. Most germanic languages speakers can read english, some german, some dutch, swedish, etc; latins can read some spanish, some italian, portuguese, french; slavs can read some czech, slovak, polish… Would be nice an option to select some “secondary” languages for our search, so for instance an austrian user could have the german locale from his browser as the most priorized language but also english, which most people can read more or less proficiently nowadays, as secondary along with dutch, danish, this way, said user would get more search results; an argentinian, besides spanish could have english, italian, portuguese, french…
This is especially useful when searching for not too common or academic topics, where unfortunately it’s probable that you don’t find all the info you would like in your language nor in english. This shouldn’t be a configuration by default but something that the users could set in Searx’s preferences.
Once again, I don’t know if is just a matter of configuring your instance to return search results in more languages than american english and the users’ or if is a task for the developers.
Regarding the quality of the results, and going back to the restaurant example, it seems that even choosing the correct language, Searx isn’t very precise when looking for local businesses and services. If I look for that greek restaurant, searchers like Google take into account my IP address, and are usually rather precise as they return results priorizing sites in my city or at least in my province. Well, I love that Searx doesn’t store my IP, but it knows it while I’m in its website. Why not to use that IP of mine that Searx temporarily knows, give me good search results based on that IP, and when I leave the site forget it forever?
Is this among the Searx engine configuration options you can tweak?
I’ve said that Searx has a lot of potential, a libre and privacy friendly meatasearcher seems a very potent tool, but not with the default values. If I really want good search results I have to go to Preferences, and, besides the language thing, add Qwant, Yandex, DDG, and some more, to the default search engines, then I get A LOT of very nice results, many often much superior to the results I get in Google. It’s amazing, but then I close the tab, the cookies for Searx are deleted and the next time I use it I’m again in the beginning, and I have to configure once more the language, search engines and other options. >_<
In Startpage, for instance, you can set your preferences and save it, at least in Firefox; no matter if you delete the cookies, caché, etc, your configuration will be there along with the other search engines that Firefox came with from “factory”.
Last but not least. When one saves a configuration an URL is provided. It’s like: http://search.disroot.org/?preferences=Ce3rE_uMQlk9K_JYjxsHJPb7nA-gxV5zJF7d… and a lot more characters. It seems to me that such a weird URL must be very uncommon, unique I’d say. So, if that URL is passed to websites is a perfect way to track us since that URL is in fact a fingerprint that nobody else on Earth has. Searx warns us when we make changes in the preferences:
specifying custom settings in the search URL can reduce privacy by leaking data to the clicked result sites
When is that URL leaked? Do websites need to know it? It should be Searx the only one that knows it, no? If it is impossible no to leak this identificative URL, couldn’t you make Searx to use more generic URLs, so many users would coincide using the same? I mean that the possible combinations aren’t infinite: if I change the default language and add a couple more of search engines, is very probable that other users may have the same configuration. Couldn’t be the same URL for all of us, so it would be a bit harder to indentify us?
If you have read the whole post I give you sincere thanks for your effort