Hi everyone,
I’m choosing a default search engine and I’m torn between DuckDuckGo and Brave Search. What matters most to me:
Privacy: what’s actually logged, and can the claims be trusted? Independence from Big Tech: DDG partly relies on Bing, while Brave has its own index. Result quality: relevance for daily, technical and niche queries. Trust: track record, controversies, business model.
Which one do you use daily, and why? Have you noticed a real quality difference? Thanks!
Why is brave even an option. Avoid that at all costs.
I stopped using DDG because it kept steering me to articles that were rehosted with paywalls by MSN instead of the actual source of the content.
I use Brave mostly, because they at least pretend to be private, and I actually get some decent results.
Unsurprisingly the best results come from Kagi, the paid service, primarily because they let you cater the results to your preferences. You can blacklist certain sites, prefer others, and even search the “small web” for blogs and such from smaller creators, and the Fediverse at large.
Why pay for some company to watch over you and your searchs? If you want real private searching it comes down to searXNG!
Agreed, hosting is probably the most private web search we have. Only your IP address is still sent to the actual search engines searXNG calls. Using a VPN would solve this last bit too.
An extreme alternative would also be Tor.
SearXNG is not absolute privacy either. They don’t each have their own index.
No one pays “for some company to watch over you”. The reason why you pay I already gave: because it’s better at what it does.
SearXNG is private iff you use someone else’s instance
Me, I like self hosting things… So it’s actually less private. Multiple search engines get each of my queries lol
Kagi in general, and if I feel extra strongly about privacy, I use Privacy Pass mode which is cryptographically guaranteed private.
I’ve been using DDG as my default for probably longer than I ever used Google, so I’ll share my thoughts on it
What I like:
- Nice UI with plenty of options
- Optional AI features
- Great image search with tons of filters and an AI slop blocker that works reasonably well
What I don’t like:
- Localized results are not good (Startpage might be better for those)
- No LLM search API, so I use Brave API for that
They’ve had controversies in the past, but I’ve not heard of any in recent years tbh
AI slop blocker
How does this work? I mean how decides it is Ai slop or not?
Known sources, digital watermarks, and user-contributed reporting.
I see. Well that makes sense and is indeed useful then.
Note, I have only seen filters for the image search. But there’s definitely a lot of slop among the images even though you select to exclude slop. There’s just so much slop out there I guess.
I view this comparable to an ad-blocker extension in the browser. It doesn’t block all ads, but it does a decent job that the result is usable. Maybe that is the effect, maybe less. As long as they do not heuristic decision and only go about known stuff, then I’m all for it. My first thought was “oh they fight ai with ai”, but that is not the case as it seems. But it depends on the sources for that filter. If they use Ai to detect, then I’m not a big fan of the filter.
Indeed. Note also that I’m only certain about the user reporting, as I’ve done some reporting of my own. I stopped after a while though. There’s so much that it feels futile.
But, the other ways of determining should be possible too.
For LLM I use open-websearch which supports a ton of search engines
SearXNG is a little tricky to set up, but worthwhile. If you’re lazy, I trust DDG over Brave any day of the week.
I went from the Google to Ecosia and then searXNG. I guess I left ecosia before they went on the big AI trip.
I use mainly Startpage and Mojeek, sometimes Ecosia and Qwant. Not so for privacy, but by far the best for image search is Yandex (yes, I know). But anyway there is always needed to use several search engines + specific page searches for an deeper research. Respect privacy there are only few conditions, the search engine don’t log your searches (avoiding filter bubble), don’t use AI of big corps, no SEO crap and don’t use tracking cookies, given in most engines except Google, Bing and Yahoo. Privacy isn’t a main problem in other engines as these and finish when you visit the searched page, where the protection from any search engine finish if you do it without proxy or VPN.
Daily reminder that the majority of alternative search engines get their upstream results from Bing or Google. As part of their agreements with Microsoft and Google, they may share some information with them. This is not a secret or anything; you can read it all in their terms of service.
Also a reminder that most of these search engines are black boxes. They’re all closed source and all we have is their “just trust us” promises of privacy and security and none of them have been audited by trusted third parties as far as I am aware. At this point, just use the one that most closely aligns with your personal values and ethics.
“Just trust us,” on the promises of privacy is pusing it, they are all openly sellling your information.
If you do any self-hosting, hosting your own instance of SearXNG is a much better way to go. No AI and you actually get results for what you search for. It pulls results from all the other search engines and you can choose which engines you want in settings, and it masks your IP so the other search engines don’t see where the requests actually come from. But even if you don’t self-host, SearXNG is just a much better privacy respecting search engine anyway.
There is a new Linux distro that includes hardened Zen browser and a tool to set up SearXNG locally. There is also an option to have the searches run through three TOR relays. It makes it all super easy to setup if you’re not someone who is self-hosting other services.
Hmmmm, interesting. What is the name of this distro, if you don’t mind me asking.
Odyssey Linux https://odysseylinux.org/
I just started self hosting a Nextcloud instance and I’m absolutely going to look into using Searx on it now. I forgot that was a possibility.
Be careful; one project turns into two, which turns into 3, which turns into a lifestyle! I use searXNG and have a wireguard instance into my LAN so that I can access the nth service I just set up that I cannot believe I have been living without when I am away from the LAN!
May I also suggest pihole be in your list of things to try out, or adguard home if that is they way you swing I will not judge. Once you get your ad blocking setup you will really start noticing the ads in other places and you will then need to go further into ad blocking, when you decide to get your own domain name may I suggest you find a provider that offers a wildcard email option, if you need a free trial service just a little longer create a new account with a new email address or if something gets spammy just have things sent to that address either bounced and deleted or just deleted before it gets to your inbox.
Be careful; one project turns into two, which turns into 3
Actually two turns into four! Otherwise I agree with your reply. :-) My next step for searXNG would be to set it up on a low powered device to access from public (for myself only), and use a VPN on it with rotating IPs. To accessing it outside my home. But I’m not ready for that yet. I’m also kind of scared to get hacked someday.
I am a dumb ass with a BA in the social sciences not computer science, but I was able to get a vanilla OpenWRT installed on a Flint 2 firewall. Sure I did a lot of reading, youtubing, and got a little AI help. The Flint 2 comes with OpenWRT installed on it but everything I read suggested putting a clean install.
I had a slow start to home labbing I started with one pihole than moved to two, than I thought about getting a VPN running but that did not really work I believe I even tried using duckDNS for that but than I did the dumbest thing anyone can do I opened port 22 in my firewall so that I could SSH in. When my internet mystiralisly stopped working correctly I called up my ISP and wow did I ever learn that opening port 22 was a bad idea for how I had it set up at the time. So that stopped me exploring homelabbing for a while, sure after I had new pi installs set up it was never project after project but I never used port 22 again (until recently now that I have crowdsec running) and never used passwordless login. But earlier this year my place flooded and I was out of my home more frequently so I needed to find a way that I could access my pihole and smb (I had given up on using nextcloud at that point but I think I am coming back around to it), so came the Flint 2. With OpenWRT the 3 or 4 projects turned into an infinite number of possibilities.
Oh man sorry to hear your place got flooded. That sucks, but good on you for not stop trying. Man also, you can’t just do a passwordless login when dealing with the internet. :D And the story with misconfiguring the port is exactly one of the things I’m scared.
NextCloud is also on my todo list,… but I’m too lazy.
Thanks but this girl has nice new flooring and a new hobby/way of life! I would disagree with you on passwordless log in, do it where ever you can but use ssh-keys and change ports away from 22.
I really do not like using ports in the 8080 range either but I think that is because of my 22 issue.
And then you find Home Assistant and don’t stop until you literally have the lights automatically be precisely on you and only you, wherever you go in your house.
I am in some self hosting lemmy group, I was given so much heck for still using ring, I needed to fix a motion sensor flood light when I figured it out and told people what I had done I got so many downvotes.
I get it ring is not good, but I have a vLAN (10.6.66.0/24) that does not have access to the WAN.
I run my containers through Tailscale to be able to access my services away from home. But yes, you’re right one container turns into adding another container and on and on. But its a better way than depending on other services that spy and track, etc. I still have a lot that I need to learn and add to the server but for the most part it accomplishes everything I want when I decided to start a server.
I use wireguard on my openWRT firewall, any ISP firewall/modem/router is a joke.
Yeah, I actually want to learn to build my own router and learn to set up openWRT, etc. Its a fairly new field to me but I know its something I need to learn. ISP routers are definitely a joke.
ISP routers are also a security issue, when that PHP thing happened I had to go through all my stuff and find what was vulnerable and either update it or replace it, luckily nothing on my firewall was affected. When sshole was the big scare I went and looked everything up on it to make sure I was not in trouble, I wasn’t I use keys with password authentication turned off, but that got me thinking I should see if there are any updates for my firewall. Turns out there were a few, I had never thought of updating a firewall before because the only times my ISP firewall would have been updated is when I broke something and the IT people would just decide to do their thing and break all of my configs, there may have been updates with that.
I saw that that sshole issue was really only an issue for people running older out of date systems, is that true?
I know I have a long road to travel learning the router, networking, security stuff but it is a road I will travel as I find time.
Yep, highly recommend it. Its easy to setup, especially if you already know a little about self-hosting.
What should someone who has got 0 idea of how to even start self hosting, start with?
I just looked up YouTube videos on using docker compose when I started. Also https://trash-guides.info/ has some great information. Services you want to use like Jellyfin for example has documents and good information for setting it up. I use docker compose so I would start with their website.
DDG has a no ai version which is what I use.
It also has a lite version - https://lite.duckduckgo.com/lite
Or the HTML version which I think looks nicer. https://html.duckduckgo.com/html/
You can add this link as a custom search engine to your browser. https://html.duckduckgo.com/html/?q=%25s
I tried NoAI DDG vs Lite vs HTML. Why do No AI and Lite have identical results while HTML is similar but might be reordered or have one or two different results?
Or the HTML version
Thanks for this! HTML version is clean af
I use qwant, mainly because its european. The quality is not that good imo, but its ok for my usage
Same, cause its European and it also is building its own independent index!
Yes. But it’s the only one, who refuses to work if you try to use it without JavaScript. So ddg for me + canvasblocker on firefox / brave browser which also blocks your fingerprint. (But since brave is a an AI and crypto company, i must say… Trust is not what i have a lot of)
What are typical server requirements for search ng? Especially storage wise?
I run searXNG on a pi 5 with a bunch of other services, it works great.
Edit: the pi 5 has a NVMe drive.
You don’t need much at all if you’re asking hardware-wise. Depends on what you want to host but I run jellyfin, searxng, slskd and many other containers on a small Intel CPU and 16 GB of ram. I run all of my containers in docker-compose and have no issues. Storage wise, all of my containers run from a 1 tb ssd, the containers themselves only store your config files, etc so you don’t need much storage for that. If you wanted to run like Jellyfin for media then you would probably want more storage.
I use Firefox with Brave search and DDG as a backup. On my iPhone, brave doesn’t always work on Firefox so it redirects to DDG. For what I researched, brave search is more private than what they getting caught for on their browser.
Show me your research. Cite your references.
Oh here’s my other comment with NYT article.
The article also has this… https://www.theverge.com/tech/1024588/brave-lawsuit-ai-scraping-new-york-times
Ok…
Ok. Give a few days to get back to my computer and look through my history. Not saying I’m correct either.
deleted by creator
Yeah I know. Didn’t saying about that though but ok.
deleted by creator
They may be more secure than others, but they been catch too many times stealing data. Privacy isn’t their strong suit.
The New York Times sues Brave web browser for AI scraping and bypassing paywalls
Firstly: The company, and especially its CEO, have a shady history which makes trusting them difficult.
Secondly: Their browser further promotes Google’s browser monpoly. For some of us that is a huge downside.
Why this hatred? And where does it come from?
The CEO donated money towards preventing gay marriage in California.
The company has some questionable businesses, for some the simple fact that its chromium based is enough.
So what the alternative ? Especially for phone i think there are no browser that block ads & tracker. Also you can’t use extensions
Canvasblocker prevents fingerprinting - which is nice. https://royfish.dev/fingerprints-are-the-new-cookies
Ironfox is a privacy focused Firefox fork
I run Fennic with all my normal Firefox extensions, I’m also pretty sure Vanadium blocks most stuftlf but I’m truly not sure.
Setup a VPN into your LAN and self host searXNG
Ironfox as mentioned and also Titanium browser which is built from Vanadium browser if you aren’t on GrapheneOS. Both block ads and allow extensions, especially the ublock origin extension.
My process is open in IronFox, if it doesn’t work open in vanadiam.
It’s a browser made by a company whose main business is advertising.
At one point they were trying a model where they adblock everything, but show you their ads instead. And then hold the money it made hostage untill the owner of the site you saw it on makes a contract with then to get paid for your ad viewing time.
But in general - an advertisement company making a browser is suspicious on so many levels…
Brendan Eich has some of the wrong ideas
