Show HN: Self-hosted, ad-free, privacy-respecting alternative to Google search | Hacker News Reader

Show HN: Self-hosted, ad-free, privacy-respecting alternative to Google search

github.com

32 pointsvinushkah42 comments

The crucial part is missing in the Readme - how it works.

Nothing else matters until this bit is explained.

From the looks of it, it's just a proxy that forwards search requests to Google. Unless these happen to go out via Tor or similar anonymizer tech, that's hardly a "privacy-respective alternative to Google search".

It does appear to strip cookies, and either some or all of Google's JS, depending the config. So, Google wouldn't see your client IP, any cookies, etc. Perhaps not fully private, but they do seem to provide better privacy.

> So, Google wouldn’t see your client IP

Yes they would. You can’t sent them a request without giving them your IP.

- If you’re self hosting it on your device, they have that device’s current public IP - If you’re self hosting at a server you have, you’re giving them that (and will be easier to track).

"Yes they would. You can’t sent them a request without giving them your IP."

I suspect you know what I meant. They can't see the IP of your browser. Rather, they see a Heroku, DO, whatever IP.

I actually tried it on my laptop.

That’s why I pointed out the 2 cases.

If that's the case, then why not just use googler [0] since it only appears to send IP address, user-agent (which is set to Google Chrome) and search term to Google [1]

Edit: Forgot to add, also supports proxies.

0: https://github.com/jarun/googler

1: https://old.reddit.com/r/linux/comments/gayf4i/googler_v41_i...

For a fully independent alternative, see https://private.sh which is uses https://gigablast.com as a data source.

Gigablast itself is an open source crawler, index, and query engine. It is a bit of a quirky one-man project, but also kind of good.

Since I found out that private.sh was owned by PrivateInternetAccess, I've been hesitant to use it given their recent acquisition by Kape Technologies.

The code for Private.sh is open source so you can actually see what it’s doing instead of trusting it.

Don’t trust, verify :)

You can never be sure that the code you see is the same as the code that powers that website. If you trust the code the best you can do is run your own server.

If you click View Source you can see that queries are absolutely encrypted. Running your own server just changes your IP but it’s still “yours.”

Finally, you shouldn’t “trust” code and run it. You absolutely need to read and understand first and finally, determine if running yourself actually brings benefit.

You can't trust code you don't understand, if you do that don't call it trust, call it faith. Your query has to get decrypted somewhere or else the server wouldn't be able to proxy the request to Google unless it is E2E encrypted but then it is no different than just using a VPN or a socks5 proxy. If that's the case better just use Tor or a VPN and have privacy everywhere instead of just on your search page.

Just to add some perspective - i'm using DDG exclusively because i'm really wary sending G any kind of traffic at all, since that will be tracked anyways.

Same goes for facebook - as long as you're logged in, you're sending them beacons for every site you visit that has a Like-button somewhere :S

> Same goes for facebook - as long as you're logged in...

Even if you're not logged in, AIUI they're still logging the IP's that load those like buttons, and anything else they can glean to match to "shadow profiles", which presumably they try to match with known profiles via cookies etc. if/when you log in later.

Use privacy badger

> Same goes for facebook - as long as you're logged in, you're sending them beacons for every site you visit that has a Like-button somewhere :S

Doesn't facebook run in its own container by default?

This happens even if you're not logged in.

You can go to the ublock origin settings and from the 'filter liste' section enable the social blocking ones. Those will disable the fb/twtr/etc. buttons completely. Or, as someone else mentioned, Privacy Badger should help without completelty disabling them.

my first thought was "if it's self hosted, don't I have to index the entire internet first before I use it..?"

.. before I read it's actually a frontend/proxy to google, not a separate thing. This is fine, but the seems a bit misleading to me.

I like the idea, I think we need more projects like this.

Minor nitpicking re: title, I don’t view it as an alternative to google search, since it still relies on google search results.

I don't think it's either minor nor nitpicking. It's not an alternative at all. If Google doesn't exist, this doesn't exist.

That said, I like the effort and it makes sense. It just shouldn't be called an alternative.

It wasn't called an alternative, you're criticizing the title of the HN post which has nothing to do with the project.

Edit: Never mind I missed the upper project description.

The project description is "Self-hosted, ad-free, privacy-respecting alternative to Google search", same as the title.

Lately I've been using searx, which I found out depends in quality greatly on the instance used. Searx.space has a list of instances you can use, and you should try another if one doesn't perform well for you.

Same here, hosting my own. It has been working great. Never looked back since i switched quite a few years back now.

I've tried plenty of instances in the past and they've all been defunct. May I ask which one you recommend?

Stinpriza.org works well for me. Tried searx.be and that one didn't give me good results.

Thanks!

Title is completely misleading

Why is this flagged? Maybe it's something HN doesn't find very useful, okay, but that's not a reason to flag it.

As someone who mostly uses HN on mobile, I occasionally accidentally flag things, and then they disappear from my feed making it impossible to correct my mistake, and there is no way (that I know of) to see a history of what I flagged. I wonder how many people accidentally flagging could trigger the notice.

reading the comments, probably because it is not what the title promises.

Yeah, I see. I still think it's a bit harsh, especially since it kinda depends on how you interpret it here (i.e. what is "Google search"? The frontend or the backend? etc.) and it was made in good faith after all.

This reminds me of a passage in the "Waterfox joins System1" blog post[1] where the maintainer of Waterfox goes into how they've recieved harrassment because some people felt Waterfox wasn't up to their standards in terms of privacy.

[1] https://www.waterfox.net/blog/waterfox-has-joined-system1/

> reading the comments, probably because it is not what the title promises.

If that were the case, admins would simply change the title.

I suspect the reason is that it's a tool that is designed to violate a specific site's TOS, which I think this tool does.

That admins could change the title doesn't mean users don't feel mislead and flag in the mean time.

I have observed that when I flag a title it is not publicly indicated to be flagged.

It's not a binary. There's a threshold of flags after which it is shown and a higher one after which it gets killed, scaled by other activity on the submission (and possibly some other internal factors, the details aren't really documented).

Perhaps a common protocol to query index search items would help with all this. New search engines pop up but how can they realistically stand a chance unless they depend om bing or google at least in the back end,even if they can they simply can't out spend or out hire google

I have been using the Russian search engine yandex.ru lately. You will get different results. It reminds me of the google of previous times with less "SEO optimized" content on-top.

I wonder what server one would need to actually self-host web search engine. And how long it would take for initial indexing?

And how much energy it'd take if everyone did that...

Addendum: Maybe something like what package managers are doing would be feasible? Download an index that's as compact as possible and search locally. Does that exist already?

This seems like something Google doesn't like, as in, they'll stop serving you results.

It's like a search engine proxy? I'm assuming flask?

Reply on news.ycombinator.com