4 ms·

While I am very willing to pay money for a user-centric search engine the requirement to do so directly conflicts with the greatest must-have feature that I nee

While I am very willing to pay money for a user-centric search engine the requirement to do so directly conflicts with the greatest must-have feature that I need from one: Anonymity. Now the issues is not just privacy but making sure it does NOT optimize search results based on prior searches or other facts it knows about me. Now the big problem is that one would think delivering the exact search results that I am looking for is a good thing. No. It is super dangerous. It can causes me to live in a bubble, blinding me of possible opposing opinions. It is feeding my ignorance. So there is a goal conflict here. I don't know how to solve it. I wish there would be something like Wikipedia for search engines but even the wiki models has its problems and biases.

memen 4y ago
I like the idea. Maybe a 'Search for me' button instead of the current 'I'm feeling lucky'. Or two columns or tabs for personalised and general results. Sounds easy, but I am not aware of any search engines that offer this, unfortunately. On the other hand, a lot of the search results bias stems from the search query itself. "Does X cause Y" for example will show results that echo the search, and not show opposite results. Simply because you did not ask the opposite or a more open/unbiased query like "X Y".
Fire-Dragon-DoL 4y ago
I thought: couldn't the details about the user live on the client? And the user could delete, modify, decide not to use them. This way you get anonimity and custom results
- Arnavion 4y ago
  That doesn't work unless the server sends all 1 million results (say) to the client up-front. Otherwise, if you prefer results from example.com but the server only sends example.com results on page 3, your client has no way of showing you those results on page 1.
  - Fire-Dragon-DoL 4y ago
    I think I've been misunderstood, in my mind the client sends all the details about themselves to the server, the server builds up the search results and send them back. Yes, the server has access to the information the client has, but it doesn't have any way to confirm that's true. It also could avoid storing that information all together if the client sends it again every time
g_p 4y ago
I think if you can assume some level of trust in your search provider (i.e. through payment - let's get to that in a minute), without which you should probably be looking for a new search engine, then you can trust the search engine to not optimise results for you. If they tell you they don't, then do, then you likely have deeper-rooted issues with the search engine you use. On the point about payments, there are some potential approaches to handling this, but they tend to be a bit complex and technical. I"m not sur ethis is really a technical problem though (or at least one where a technical solution is what's needed). A challenge of taking traditional payments is that the identity of the payer is inherently "linked" (to some extent) to the transaction. You could maybe use a privacy.com type intermediary or similar. What you could do however, is buy a "kagi voucher" or similar, which would be an RSA-blind-signed token that you generated client-side in JS. You'd send it to the server during the payment transaction, and be given back a signed version of the token, which you unblind client-side, then give a backup copy to the user to save, and set it as a cookie to authenticate payment has been made. That should, in theory, at least un-link the payment from the user. Ultimately though, a bad actor can still trivially correlate your searches (and payment, if you used same IP or browser fingerprint), and all your searches are linked by a common cookie. So you decide to create some kind of token fountain, which you can auth to with a blind-signed token, and get 50 "search passes" from. Those could be unique per-search, and just be "bearer tokens" you can use. You'd have to trust that this token fountain doesn't record which search passes were issued to which "blind token". Ultimately, it feels like this problem then becomes "do you trust your search engine?" - I'm not sure this is a problem that technology needs to solve, as ultimately your search provider is seeing your queries and IP, and most users won't realistically change this. A provider that isn't incentized to act against your interests (i.e. something like Kagi which isn't advertising to you or selling data) seems to address this for most user scenarios.
- cardanome 4y ago
  I don't think there is much cause for worry that Kagi themselves might use the data for anything bad BUT data leaks do happen and more importantly Kagi will need to play ball with whatever government that exists in the markets they want to have users in, be it the USA but also China, Russia or whatever. We already know that Google is in fact cooperating with these entities. Not only by giving out user data but also by actively censoring certain things. So there is a precedent.
libraryatnight 4y ago
It sounds like you use the internet way different than me. I can't remember the last time I searched something where the end result wasn't mostly objective. Documentation, directions, tour information, movie times, programming questions, patch notes. Maybe seek contrary views out in the world. The internet is a terrible place to form a world view and to really communicate in general.
- t0bia_s 4y ago
  Agree, however we are already in trap of algorithms. Look how few years of using social networks spread ideologies that would be laughable back then, ie "there is more than two genders" or "earth is flat".
wruza 4y ago
It can causes me to live in a bubble, blinding me of possible opposing opinions. It is feeding my ignorance. This made internet completely unusable for discovery. A big shiny window into another tiny room. Absolute BS. Recently I discussed it here on HN and was advised to like, dislike, subscribe and so on on youtube (things I rarely have done on my old account) to break out of this bubble. Well, it doesn’t work. It just suggests the content I immediately dislike or ban entire channels and it all settles to the same echo chamber as before.
- kqr 4y ago
  I think many platforms optimise for interactions, with no regard for whether they are positive or negative.
  - wruza 4y ago
    As a result, I often find myself closing youtube tab because there is not only nothing to watch, but nothing to scroll through even. On a platform with billions of videos, at least thousands of which I might like to watch if given a chance. I find this incredibly stupid. Same for google search, really. It turned from “I’ll find you that if it exists” into “Look, if you change your query to ‘cool products if at all barely relevant to …’, I have literally billions of results for that”.

memen 4y ago

I like the idea. Maybe a 'Search for me' button instead of the current 'I'm feeling lucky'. Or two columns or tabs for personalised and general results. Sounds easy, but I am not aware of any search engines that offer this, unfortunately. On the other hand, a lot of the search results bias stems from the search query itself. "Does X cause Y" for example will show results that echo the search, and not show opposite results. Simply because you did not ask the opposite or a more open/unbiased query like "X Y".

Fire-Dragon-DoL 4y ago

I thought: couldn't the details about the user live on the client? And the user could delete, modify, decide not to use them. This way you get anonimity and custom results

Arnavion 4y ago
That doesn't work unless the server sends all 1 million results (say) to the client up-front. Otherwise, if you prefer results from example.com but the server only sends example.com results on page 3, your client has no way of showing you those results on page 1.
- Fire-Dragon-DoL 4y ago
  I think I've been misunderstood, in my mind the client sends all the details about themselves to the server, the server builds up the search results and send them back. Yes, the server has access to the information the client has, but it doesn't have any way to confirm that's true. It also could avoid storing that information all together if the client sends it again every time

g_p 4y ago

I think if you can assume some level of trust in your search provider (i.e. through payment - let's get to that in a minute), without which you should probably be looking for a new search engine, then you can trust the search engine to not optimise results for you. If they tell you they don't, then do, then you likely have deeper-rooted issues with the search engine you use. On the point about payments, there are some potential approaches to handling this, but they tend to be a bit complex and technical. I"m not sur ethis is really a technical problem though (or at least one where a technical solution is what's needed). A challenge of taking traditional payments is that the identity of the payer is inherently "linked" (to some extent) to the transaction. You could maybe use a privacy.com type intermediary or similar. What you could do however, is buy a "kagi voucher" or similar, which would be an RSA-blind-signed token that you generated client-side in JS. You'd send it to the server during the payment transaction, and be given back a signed version of the token, which you unblind client-side, then give a backup copy to the user to save, and set it as a cookie to authenticate payment has been made. That should, in theory, at least un-link the payment from the user. Ultimately though, a bad actor can still trivially correlate your searches (and payment, if you used same IP or browser fingerprint), and all your searches are linked by a common cookie. So you decide to create some kind of token fountain, which you can auth to with a blind-signed token, and get 50 "search passes" from. Those could be unique per-search, and just be "bearer tokens" you can use. You'd have to trust that this token fountain doesn't record which search passes were issued to which "blind token". Ultimately, it feels like this problem then becomes "do you trust your search engine?" - I'm not sure this is a problem that technology needs to solve, as ultimately your search provider is seeing your queries and IP, and most users won't realistically change this. A provider that isn't incentized to act against your interests (i.e. something like Kagi which isn't advertising to you or selling data) seems to address this for most user scenarios.

cardanome 4y ago
I don't think there is much cause for worry that Kagi themselves might use the data for anything bad BUT data leaks do happen and more importantly Kagi will need to play ball with whatever government that exists in the markets they want to have users in, be it the USA but also China, Russia or whatever. We already know that Google is in fact cooperating with these entities. Not only by giving out user data but also by actively censoring certain things. So there is a precedent.

libraryatnight 4y ago

It sounds like you use the internet way different than me. I can't remember the last time I searched something where the end result wasn't mostly objective. Documentation, directions, tour information, movie times, programming questions, patch notes. Maybe seek contrary views out in the world. The internet is a terrible place to form a world view and to really communicate in general.

t0bia_s 4y ago
Agree, however we are already in trap of algorithms. Look how few years of using social networks spread ideologies that would be laughable back then, ie "there is more than two genders" or "earth is flat".

wruza 4y ago

It can causes me to live in a bubble, blinding me of possible opposing opinions. It is feeding my ignorance. This made internet completely unusable for discovery. A big shiny window into another tiny room. Absolute BS. Recently I discussed it here on HN and was advised to like, dislike, subscribe and so on on youtube (things I rarely have done on my old account) to break out of this bubble. Well, it doesn’t work. It just suggests the content I immediately dislike or ban entire channels and it all settles to the same echo chamber as before.

kqr 4y ago
I think many platforms optimise for interactions, with no regard for whether they are positive or negative.
- wruza 4y ago
  As a result, I often find myself closing youtube tab because there is not only nothing to watch, but nothing to scroll through even. On a platform with billions of videos, at least thousands of which I might like to watch if given a chance. I find this incredibly stupid. Same for google search, really. It turned from “I’ll find you that if it exists” into “Look, if you change your query to ‘cool products if at all barely relevant to …’, I have literally billions of results for that”.