SearX Most of us will use these every day (usually several times, in fact), so it's important to carefully consider your selection of choices. Not only do we want great results, but - since we will often share some really sensitive data with them (such as medical, travel, or even that we're interested in loli or shoplifting) - they better not be doing dirty stuff with it. There are also various additional functions that we'd like our search engines to provide. Now why did I use the phrase selection of choices? Because - as you will see - the situation with search engines is not as good as with E-mail providers, so we're forced to switch around since there isn't a single one that has everything we need. Without further ado, let us check what is available:
A while after finishing this article I've had a "revelation" - the situation with search engines is even worse than I thought before. Aside from Wiby (which has very few searchable sites in it) and Mojeek, the search engines mentioned here depend on others for their results. By "others" I mean either Google, Bing or Yahoo - three massive privacy violators. When you search through, say, Swisscows - the query is first sent to Swisscows, which forwards it to Bing, which then returns the result to Swisscows and finally to you. There are several problems with such a setup:
This is not that different from using Invidious, Nitter, Facebook Container or custom Discord clients for their respective services - the violator still calls the shots. And we - like the wives dependent on their abusive husbands - keep coming back for more in hopes we can prevent some of the damage. But - as we rely more and more on the violators - it gets harder to dump them. Keep this in mind while reading through the below recommendations.
SearX UPDATE June 2022: rewritten. SearX allows you to choose the search engines used, and proxies your queries to them through itself. It has no own index so all the issues from the relevant section still apply. Still probably the best option to use these days, if you find a good instance. The level of support for different search engines can vary heavily. Bing, in particular, doesn't seem to work anywhere, and all other options aside from Mojeek (which will almost always block SearX) and Gigablast, censor. Yandex is not able to be proxied by SearX, either. Still, most of the relevant search engines are supported. So you might be wondering why use anything other than SearX. Well, it's because this happens more often than not:
However, using SearX instead of the engines directly obviously provides superior privacy, not giving the proxied engines your IP or loading their trackers. Remember, though, that instances could still choose to collect data (read privacy policies if they exist). The best option is to spread your searching over several instances. SearX provides very useful Web Archive integration as well, helping avoid Cloudflare. Onion domains are possible if the instance has installed them. Searching itself works without JavaScript, even with images. NG has seemingly improved the image results, as well. I recommend using one instance for images and another for regular results; this allows adding both as different shortcuts to the address bar, and without getting image results stuck in between your regular ones.
Brave Search UPDATE: I don't want to call it "decent" anymore. Even though it apparently has its own index (archive) (MozArchive), it clearly deranks alternative content in favor of mainstream trash (results similar to Google's), even if the latter is barely relevant to what you searched for. That's in contrast to DuckDuckGo or Qwant, whose results are more in line with what you've actually typed. Edit: I was too lenient. The censorship is exactly like Google's - meaning you will never find the truth about the moon landing or COVID vaccines there even if the query asks for exactly that. What you will find - though - is a bunch of irrelevant "fact check" or "science loving" sites, or ones shitting on "conspiracy theorists". I am now doubting if they truly have their own index instead of lifting everything from Google - unless they chose to index only the evil sites and ignore all the good ones, which would be even worse. They lie about it, BTW:
No, Brave Search does not filter, downrank, or censor search results. Nor will we change our search algorithm to increase or decrease the prominence of results in response to current events or anyone’s political, religious, ethical, or other beliefs. Brave Search—like Brave itself—is intended to be a user-first portal to the web, free of Big Tech’s manipulation.
But everyone can easily verify the censorship for themselves, with a query such as "9 / 11 was done by the government". And just with that, this brainwashing facility pretending to be a search engine should be trashed. UPDATE June 2023: if you want proof, here are the first 6 results for Christchurch shooting was faked
- my favorite query for testing these kinds of things:
Five of them attack conspiracy theories, maybe worse than what even Google does. And that's not everything that's bad about Brave. Though its privacy policy (archive) (MozArchive) seems to say that the IP addresses are not stored:
You can search for ‘cafes near me' and Brave Search will use the IP address broadcast by your device but without sharing that IP address and without storing it. This means we can't tie your searches to you and your location.
It doesn't say literally anything else. We have to go to another page (MozArchive) to see just how much stuff Brave stores:
- Number of daily/weekly/monthly visits
- Number of returning visits
- Number of search queries per day
- Average query length
- What percentage of queries led to a user clicking a search result
- How many users have chosen to leave feedback about Brave Search
- The operating systems people use when they visit (e.g. macOS, Windows, etc)
- The browser you’ve visited from (e.g. Brave, Chrome, Safari, etc)
They say this is all opt-in
- but in reality, not only is the tracking opt-out (enabled by default) but you cannot even change it without JS and not at all in Pale Moon. This means they lied in their privacy policy by pretending that they will only collect the data if you allow them
, which sounds as if it requires actual user action to happen. As usual, cynicism wins over hope. This should have been expected - it's still fucking Brave, and there has been a lot of shady shit going on with them. How do we know they won't include some sponsored results or add referral links, the way they did with their browser? How do we know they won't eventually partner with a violator, like they did with Uphold? And that's the biggest flaw of this search engine. Due to Brave's reputation taking several blows to the chin already, we just can't trust it anymore. Image search does not work without JS. UPDATE June 2022: Text queries used to work through TOR without JS, but now they bring up this:
BS requires JS and WebAssembly enabled to search through TOR, enforcing the usage of bloated browsers. See? I told you to not be hopeful. Trash it and don't look back. UPDATE: there is now an onion, so I guess they do not hate TOR now at least on the onion.
Swisscows UPDATE: it's worse than I thought - keep reading! If we only took into account the privacy policies (archive) (MozArchive), this little known search engine from the Swiss Alps would rank very highly indeed:
We do not collect any of our visitors’ personal information. None whatsoever. When using Swisscows neither your IP address is recorded nor is the browser you are using (Internet Explorer, Safari, Firefox, Chrome, etc.) collected. No analyses are made, which operating system our users use (Windows, Mac, Linux, etc.); your search are not recorded either. We record absolutely no data from our visitors. The only information we store is the number of search requests entered daily at Swisscows, to measure the total overall traffic on our website and to evaluate a breakdown of this traffic by language and mere overall statistics.
No IP or browser data is stored! Swisscows even realizes (unlike, say, DuckDuckGo) that saving the search queries is also dangerous, even if allegedly disassociated from so-called personal information:
Furthermore, it is important not to store any search terms, given that these can also contain personal data. (Just think about someone who enters their own name and/or their insurance number into the search box.)
For a while I thought that - while they might not store anything themselves - they might still share some data with Bing (which they use for their results) - but my suspicions were defused:
Swisscows does not transmit any personal information to third party search engines or to the providers of sponsored results.
All of this is according to their privacy policy only. However, we know better than to believe a privacy policy blindly - especially when there are some problems with the trustworthiness of Swisscows. First of all, they require JavaScript to display the page at all - which is a terrible design; worse yet - the code is heavily obfuscated - almost as if they were hiding something. I got a report of Swisscows including redirect links in its results. At first I thought it's some mistake, but decided to investigate and confirmed it. They look like this:
http://www.smartredirect.de/redir/clickGate.php?u=[UniqueUserId?]&m=12&p=[YetAnotherID?]&q=[YourSearchQuery]&url=[TargetURL]&r=https%3A%2F%2Fswisscows.ch%3Fquery%3D[YourQuery]
Shady as fuck and contrary to their privacy policy of not sending your data anywhere - because that redirect service will absolutely get your browsing history with what seems to be an user ID (unconfirmed yet) attached to it; as well as the search engine you used (in this case, Swisscows - but who knows what other partners in crime they have). According to this stat site (archive) (MozArchive), Smart Redirect seems to collect the browsing history of almost 20K different people every day - a pretty significant violator, then. What this service actually does is show sponsored results that are indistinguishable from the regular ones aside from a little image next to them (for which Swisscows makes another third-party request). If you click on those links, your browser will make a request to an affiliate marketing company such as awin or linksynergy - which also seem to contain unique user IDs. Sometimes, searches from the Tor network will be denied - this might depend on which exit node you get. Regardless, they lied about not sending your data to third parties - safe to say, then, that the myth of the little private Swiss Cow is dead and buried. But let's dig a little deeper:
The only legitimate reason for our beginning to collect personal data would be the existence of a legal warrant or a court order, which would require us to do so in connection with a specific user, who is suspected of such a serious crime that such a violation of his or her privacy would be justified.
So they will perform targeted surveillance without notifying you when the government comes knocking (that's the great Swiss Privacy Law in action). The search engine also has this "family-friendly" shtick, which means you might get a message such as:
Dear user, the entered word is not allowed for under 18 year olds, since we have decided on the protection of minors, the word "insert search term" is excluded from the search. Thank you very much for your understanding!
This is activated by many porn-related queries and some violence-related ones but you can still find really graphic stuff like people's limbs being cut off pretty easily. And thankfully, information alone does not seem to be censored.
Swisscows also has a semantic map
feature which displays additional terms relevant to the search you've typed, narrowing the results to what you might be looking for. I haven't found
it very useful though, or maybe I just don't get how to use it. The search engine does not run without JavaScript enabled at all (can only see a white background - bad design!). XHR is also
required to display the actual results (which come from Bing exclusively). Used to not work in Pale Moon but that seems to have been fixed recently (whether by them or by PM...). Swisscows can
search for images, videos, and news; it also includes the only (AFAIK) privacy-aware translator which is unfortunately also family-friendly (cannot translate swear words...) does not
include the translator anymore, so the one thing it was useful for is gone. Of course, the Swisscows Privacy Wall came crashing down above; it's not a sure thing we can even trust its policy
when they've already broken it. So, Bing might be getting your data after all. Having problems with Tor seals the deal even further. Another one bites the dust, as they say!
StartPage A privacy-based proxy for Google search. Claims to believe privacy is a fundamental human right
; let's see how much do they actually follow that belief:
Why we don’t collect any "personal data"
We saw the perils of that in DDG's section - where, for example, measuring engagement of specific events on the page
was considered non-personal, along with the actual search
queries. But let's give the benefit of doubt and see what is StartPage's interpretation of the term:
We don’t record your IP address
With the most important potential roadblock out of the way, the privacy train is running at full speed!
We don’t serve any tracking or identifying cookies
No hiccups so far.
We do measure overall traffic numbers and some other – strictly anonymous – statistics. These stats may include the number of times our service is accessed by a certain operating system, a type of browser, a language, etc.
The privacy train is slowing down. I spoke of the problems with so-called anonymous data in DDG's section. Remember: the only anonymous data is no data
. But it gets worse. StartPage
includes advertisements from Google (these cannot be removed by uMatrix, only element hiding) on top of their search results page, and:
In order to enable the prevention of click fraud, some non-identifying system information is shared
Aaand the privacy train has been totally derailed! Now you have to trust StartPage's determination of what is non-identifying
enough to be safe in Google's hands. Now they don't say
exactly what that data consists of - but if it's the same set mentioned above that StartPage uses for their statistics, it would definitely leave the possibility of revealing someone thanks to
browser fingerprinting. So StartPage sends potentially identifying data to Google. It's not all that bad though - at least they apparently don't store it themselves. And then there's this:
Any request will have to come from Dutch judicial authorities. We’ll only comply if we’re legally obliged to do so. But we’re not likely to receive requests by governments to hand over user data – simply because we don’t have any.
And:
We will never comply with any voluntary surveillance program
So it seems that, at least - you're pretty safe from the government's prying eyes. The other issue with StartPage are the absolutely terrible search results. That is
because Google runs a massive
censorship campaign (archive) (MozArchive), deranking any alternative or conspiratorial content. This means you will even get results that don't relate
to your query at all - and I've confirmed it with tests. Try searching for something like "was the Christchurch shooting faked" and you will see what I mean. Then compare with Qwant
or DDG. I did this for many other queries which would result in displaying alternative or conspiratorial websites (if the search engine was honest, that is), but instead showed me irrelevant
trash. Also, just like Google, it supports no other ways to search aside from Web and Images. StartPage works perfectly with JavaScript disabled and has the handy Anonymous View
feature, allowing you to visit returned sites without revealing yourself to them (however, much functionality will be disabled).
Interestingly, a long time ago StartPage used to run a search engine called IxQuick that used their own index. I used that extensively when it was around and the results were pretty good
AFAIK. Why did they take it down and submitted completely to the Google botnet? We could have had an actual privacy-based search engine with no dependency on the tech giants
and no censorship (since StartPage doesn't seem to believe in it, unlike Qwant). But with the situation as is, I cannot recommend StartPage at all due to the censored results
and sharing your system data with Google. Use only for emergencies when you really need Google's huge web index. UPDATE: actually, fuck that. StartPage
just literally got bought by an advertising company (archive) (MozArchive) and also ran a Mozilla-esque PR piece (archive) (MozArchive) defending the acquisition. UPDATE
November 2020: now they became totally authoritarian, hating TOR, VPNs and even private browsers
:
This reminds me of malicious Cloudflare browser checks, where they compare your setup to what they consider "safe" to decide whether to let you through. I found this out in November, but it's
probably been around since earlier. To be honest, I've always felt something was off about these guys; their image was "too clean" if you know what I mean. Now we have proof that for them, privacy
has only served as a useful slogan; they never actually gave a shit about it. Therefore, I can honestly advise to avoid these frauds completely now!
UPDATE March 2022: Perhaps I wasn't clear enough, so let me make a stronger statement. StartPage's (Google's) censorship is absolutely disgusting. You will find nothing but what your malicious rulers want you to find. No flat earth material, no COVID skepticism, no analyses of staged events like the so-called US capitol riot. StartPage will ignore your search query if it would result in unapproved opinions, it is nothing but a drip of propaganda. If you like being a slave, sure, keep using StartPage. But I'm a free man, and I throw mind controllers like StartPage into the trash!
Wiby.me An interesting new search engine dedicated to lean, personal, old-school websites. Really recommend keeping an eye on that one - and you can literally support the creation of a better Internet by submitting websites to it (if you know a good one, you should really do that). No big corpos allowed here! Does not require JavaScript (actually, there is not a single script on its page). Keeps logs for 48 hours (archive) (MozArchive), though.
DuckDuckGo The search engine that doesn't track you!
Or at least, that is what it claims. The marketing is slick and I must admit I fell for it initially and was a fan for a long time - but to
be honest, there were always red flags. DuckDuckGo hosts a privacy site (archive) with some great guides. They've refuted (archive) (MozArchive) myths (archive) (MozArchive) and went after giants
like Google (archive) (MozArchive) - always a plus in my mind and a huge reason to consider a provider trustworthy. Their Twitter account posts privacy stuff all the time, etc. At a glance, DDG appears as just a
group of people like you and me, looking to protect their privacy and creating a service to do so (unlike Mozilla's in-your-face manipulation). Unfortunately, like the great magicians they are,
they've made a really convincing illusion - but still an illusion; and I will show just how.
As I said, the red flags were always there. Gabriel Weinberg - the DuckDuckGo founder - used to run the Names Database (MozArchive), a social network allowing you to reconnect with your old friends from school. However, the most important feature - sending messages - was locked behind a paywall which could be bypassed if you invited 24 people to the network. They claimed that you could remove yourself from the database if you wanted to:
We enable people to remove themselves from The Names Database at any time, which instantly and automatically deletes any personal information associated with the profile removed from The Names Database.
However, the service did not really respect the user (MozArchive):
(e) The Terms may be modified only by Company posting changes to the Terms on Website. Each time User accesses Website, User will be deemed to have accepted any such changes in effect at the time of access.
Just by visiting the website you've accepted all its terms (what if they've included a line "we can kill you in your sleep") while you weren't looking? Haha.
(a) Company owns all information submitted to it, regardless of who or what submits it, what is submitted, or how or why it is submitted. Such information includes but is not limited to any and all information submitted by User or about User and any User Information submitted during membership or through the normal use of Website and the services available through it.
(b) Any successors or assignees of Company will by default acquire for their own use, in accordance with the Terms, all information collected by Company, including but not limited to all information associated with Website as specified in section 2(a) above.
So everything you submitted there ceased to become yours (or even if someone else posted your personal info), and could have been transferred to any other company. Later versions of the agreement (MozArchive) contained even more egregious stuff such as:
Opobox reserves the right (but not the obligation) to remove or edit Your Information
you may NOT [...] (e) "meta-search" any Opobox Web Site; (f) forge headers or otherwise manipulate identifiers when communicating in any way with the Opobox Web sites [...] (j) use data mining or any data gathering or extraction tools; (k) copy, reproduce, modify, create derivative works from, distribute or publicly display any content (except for Your Information) from the Opobox Web Sites;
So I can't change my user agent; use wget, httrack or curl; or even take a screenshot of the site. Funny shit. And this is all stuff that the later DDG founder agreed to. Then he sold the database to another company, Classmates.com (archive) (MozArchive), which added this to the agreement:
By registering on the Opobox Web Sites, as an added benefit you will also be automatically registered on Classmates.com, which is owned and operated by Classmates Online, Inc. ("Classmates"), Opobox's parent corporation. To complete this registration on Classmates.com, you consent to have Your Information provided to Classmates by Opobox. Classmates services are provided pursuant to its own Terms of Service and Privacy Policy.
So the privacy guru that later founded DuckDuckGo sold all your data to an even more unethical company (archive) (MozArchive). Did he have a sudden change of heart and created the uber-privacy-respecting DDG? Let's check it out:
DuckDuckGo used to claim in its privacy policy that no cookies are used by default, but some years ago it turned out they were setting a cookie from a third party (MozArchive), contrary to their policy. It was only on their help page and they fixed it quite fast. So is that all I've got? A cookie from ages ago?
In their Privacy Policy (archive) (MozArchive) they proudly
proclaim, in big letters, that they don't collect or share personal information
. The big question here, of course, is what is meant by personal information
. It turns out that,
for DuckDuckGo at least, search queries alone are not personal - even if you search for something only you could know. I'm pretty sure most people would disagree with that.
DuckDuckGo claims that it is fine if you don't store the user agent or IP address along with the search, though:
We also save searches, but again, not in a personally identifiable way, as we do not store IP addresses or unique User agent strings. We use aggregate, non-personal search data to improve things like misspellings.
What this non-personal search data
consists of, is of course not stated. Is it really only the search queries DDG saves? We do know that, for example, DuckDuckGo tracks Pale
Moon usage through a parameter in the search query (?t=palemoon
), which is there by default. This also likely happens for their other partners (archive) (MozArchive). Is that not already personal? Could it not be used to link your searches? The funny thing is, DuckDuckGo has spent much of their privacy policy criticizing Amazon for leaking their search queries (archive) (MozArchive), which
were able to be linked to single users because they fucked up their anonymization. Can you ensure this also couldn't be done for DuckDuckGo's saved searches? After all they do track other
information as I've shown earlier. As DDG itself says (archive) (MozArchive), The only truly anonymized data is no data
- so why not just not store the search queries? From the techcrunch article:
The most serious problem is the fact that many people often search on their own name, or those of their friends and family, to see what information is available about them on the net. Combine these ego searches with porn queries and you have a serious embarrassment. Combine them with “buy ecstasy” and you have evidence of a crime. Combine it with an address, social security number, etc., and you have an identity theft waiting to happen. The possibilities are endless.
How long does DDG store the search queries? That is not explained! They do admit they will comply with court ordered legal requests
. What could law enforcement do if they came across
queries that only I could have searched for? Could those be used against me? No idea but I'd much rather not have that happen! Then, DuckDuckGo admits to running experiments on their users (archive) (MozArchive):
First, you might notice that when you search DuckDuckGo, there may be an "&atb=" URL parameter in the web address at the top of your browser. This parameter allows us to anonymously A/B (split) test product changes we make to DuckDuckGo.
Second, we measure engagement of specific events on the page (e.g. when a misspelling message is displayed, and when it is clicked). This allows us to run experiments where we can test different misspelling messages and use CTR (click through rate) to determine the message's efficacy.
For example, our browser extensions and mobile apps will send an atb.js request with each search made. These requests allow us to count approximately how many devices accessed DuckDuckGo
That's quite a lot of information, and right after criticizing so-called anonymous data collection! This is exactly what I've criticized Mozilla for doing and in fact what prompted me to rewrite this whole section (and article). Gabriel Weinberg has shown himself to be hypocritical and broke trust on many occasions. For a recent one, he has put his shirt selling service behind Cloudflare (a MITM) without notifying the buyer in any way. So what's the final verdict on DDG?
There's no IP storage nor most of the other data search providers usually collect. Third party requests are not made. There
is a JS-free version (results don't go beyond first page) and they do host a Tor hidden service (which blocks non-TBB
browsers - tested with Iridium). For the non-privacy relevant stuff: the search results are really fucking good; Bing and Yahoo are used for them, as well as their own crawler. More
than that - DDG displays "instant answers" from over than 400 different sources in the top right corner of the window. Has ads that can be disabled in the settings. In addition to regular
sites, DDG can search for images, videos (YouTube results only) as well as news. However, digging deeper shows several serious issues with the founder's lack of reliability and trustworthiness.
He clearly does not truly care about you or your privacy - as the recent Cloudflare inclusion shows. Being the great businessman that he is, he has pretended very effectively
that he does and took great advantage of the recent privacy scares. So, if you care only about pure privacy / functionality, DDG is pretty good still - but due to the shaky ethical foundation
and enough cracks if you look hard enough, I cannot recommend it as enthusiastically as I did before. You could say it's the Tutanota of search engines - good enough as an
entry point, but you can do better. There is nothing out there even approaching the level of RiseUp's email service, though; so DuckDuckGo remains a viable option for regular usage still. UPDATE September 28, 2021 Now blocks Indian IPs due to some IT Act 2000 (presumably a law). UPDATE March 2022: they now censor (MozArchive) Russian disinformation
. Fuck them.
UPDATE June 2022: DuckDuckGo now includes a misinformation detector:
Are they going to remind me to take my booster next? Earlier I have said DDG belongs in high tier despite the privacy and historical issues. But with the recent violations, CuckCuckGo belongs in the trash instead.
Qwant UPDATE September 2021: French search engine using its own index plus Bing. As far as I can see, most of its flaws have been fixed. TOR and VPNs are accepted now. The search results are not being "eaten" anymore. JavaScript-free for all the categories, with a better interface than SearX. The side options have also been improved, with more sources for videos (Dailymotion in addition to YouTube) and news (but the social media and music categories have been deleted). Qwant Lite has a massive flaw, though - namely that you can't copy the link addresses since Qwant sends them all through some kind of local redirect. This makes it annoying when you just want to link something in an article instead of going there yourself.
However, the privacy policy has been revised (MozArchive), and it's become worse. I suspect they are simply revealing what was the case all along. Anyway, let me analyze the updated privacy policy:
Qwant stores your search queries for 12 months, and for the first 7 days connects them to an unique identifier:
Qwant retains for 7 days the keyword(s) entered associated with a pseudonym identifier calculated from the User Agent of your browser and the salted hash of your IP address. After this period, the keywords are no longer associated to an identifier and are retained for 12 months for aggregate statistical analysis
Microsoft (this partner
) gets your full IP and stores it for 18 months:
Qwant may also collect and transfer to this partner your full IP address.
This data is transferred to this partner within the European Union, and may be retained in accordance with Bing’s Privacy Policy for a maximum period of 18 months.
Qwant, also, collects pretty much everything in order to evaluate its audience
:
- – The source of the user’s visit to the site (the HTTP referrer and the search field used to trigger the request);
- – Information about the type of link clicked (for example the position of the clicked link in the results with the number of results proposed);
- – Some information deduced from your User Agent (type of browser, operating system, major version of the browser,…)
I can't say I'm very impressed with their privacy policy, regardless of their front page posturing. Especially the Microsoft partnership is worrying. Anyway, let's check out their ToS (archive) (MozArchive):
In the event that you notice content that may relate to apologies for crimes against humanity, provocation to or apologies for acts of terrorism, incitement to racial hatred, towards people on the basis of their sex, their sexual orientation or identity or handicap, child pornography, incitement to violence, attacks on human dignity, when browsing the Services, you have the option to let us know by contacting us at the address legal@qwant.com (also available at our FAQ).
All I need to know is here. Qwant accepts the removal of entries that violate their favorite victimhood issues. Wouldn't be surprised if they are also covid cucks who will remove alternative information about it.
What is the verdict, then? Privacy policy isn't that great anymore (Microsoft is getting your shit), but otherwise it's fine. The censorship is worrying but I haven't heard a practical case of it. UPDATE September 28, 2021: they now block Indian IPs since about a week ago. Gave a non-answer (MozArchive) when asked about it:
Hello, this is a difficult decision, but we have decided to close access to our services in some countries where we feel we cannot provide the expected quality of service. We apologise for this. Best regards.
UPDATE March 2022: another bunch of cucks that censor so-called (MozArchive) war propaganda
(but only if it comes from Russia). Fuck them too.
MetaGer This one from Germany also advertises itself as private, but it not only stores your IP (archive) (MozArchive)...
For this purpose alone, we store the full IP address and a timestamp for a maximum of 96 hours
...but also shares a part of it with advertisers:
To receive this advertising, we give the first two blocks of the IP in connection with parts of the so-called user agent to our advertising partners.
Then, their website collects and stores the following data for up to one week:
Your IP-Address, Name and URL of the retrieved file, Date and time of access, The referrer you sent, The user agent you sent
So, as we can see, MetaGer is not so great for privacy (Edit: but it's not so bad either. At least we get the duration - which most services don't bother to give - and it's not long). A later section says this:
When using the MetaGer plugin, the following data is generated:
IP-Adress: Will not be stored or shared.
User-Agent: Will not be stored or shared.
It would be easy to assume that searching through the MetaGer plugin (unlike their site) stores no logs, but that would be naive and wrong. In fact, the first sentence of the Accumulating
data by context
section disconfirms this interpretation:
When using our web search engine MetaGer via their web form or through their OpenSearch interface, the following data is generated:
And then comes the stuff about IP storage and everything. So, all the above means is that the plugin usage stores no additional data over what the site itself already does. So, MetaGer's privacy is not that good, but at least you do get a maps service that stores no logs:
When using the MetaGer map service, the following data is generated:
IP-Adress: Will not be stored or shared.
User-Agent: Will not be stored or shared.
Search query: Will not be stored or shared.
Location data: Will not be stored or shared.
UPDATE March 2022: I don't know why I ever said the search results aren't that great, because they are. I recommend using either the onion or the German mirror, as they include Yandex in their searchable index. This will find you some results the English version won't, though the onion will throw you a captcha sometimes. There is a great Open Anonymously
feature, which means you can view the sites through MetaGer, instead of going to them directly (this is the same as StartPage's Anonymous View
). MetaGer, also, has pretty good image results - which work with JS disabled, as everything else other than the maps. The privacy policy isn't great, but at least it's concise and contains no unnecessary posturing, which is commendable, I guess. You can change the language used for the results in the hamburger menu at the top, last option.
UPDATE August 2023: MetaGer, instead of just the Bing ads at the top, now also serves a bunch of "partner" results mixed in with the others (no idea when they've implemented it, but seems pretty recent). When you click them, you are taken to the destination only through an intermediary site tatrck, which is Cloudflared - so for TOR users, or users of unconventional browsers etc. those results will be unavailable. You also can't copy a "partner" search result and share it with someone without that intermediary getting in the way. This makes MetaGer now unsuitable for its main purpose, that is actually finding information and being able to read it. We won't even get into the data collection done by tatrck itself; it being Cloudflared indicts it enough. Edit: way, way too many sites are partners
, which also disables the Open anonymously
functionality for them, that I was hoping to use for some CFed sites. I think MetaGer just killed itself.
Another one with seemingly zero logging - we don't store your search terms, ip address nor information about your browser.
- and even has a Tor domain. Supports searching only for regular sites (results sourced from Yandex - very high quality
according to my short testing) and images (on the other hand, these suck). Provides Instant Answers similar to DDG and SearX, as well as proxy links (with JS stripped out to
prevent tracking and deanonymization). As with Ecosia, MetaGer and StartPage - relying on a single source for the results, regardless of its quality, is subject to eventual bias and censorship.
That, as well as absolutely zero customizability (can't change the basic, ugly default theme, or even turn off autocomplete) means it's probably worse than DDG. Edit: bumping it down since the
amount of results is being cut off for some reason. Apparently the engine itself contains much more functionality - but the actual instance sucks, and that is what I'm rating.
Mojeek I'm just going to rewrite this. In an age of widespread (archive) (MozArchive) censorship (MozArchive), Mojeek is significant because it is focused on unbiased search results (archive) (MozArchive):
Mojeek doesn't have an agenda, and neither does it support any particular view, whether that's in regard to politics or if blue is a nicer colour than green. And that's the way we believe search engines should be. Our task is to provide you with relevant search results, not to force a point of view on you, or hide ones we don't agree with.
As far as I know, no other search engine makes this commitment. Mojeek can do so because it is using their own index, instead of relying on the censoring violators. Though the results are not as good in terms of e.g scientific or technical queries, they're a lot better than I remember them being a few years ago. And - in an age of mind control - a hard stance against censorship is more important than slightly weaker results. JavaScript is not required at all, even for images. What about their privacy (archive)? They don't store IPs:
IP addresses are not recorded, instead the IP address is replaced with a simple two letter code indicating the visitors country of origin.
There is also no third party sharing:
The remaining log data (country, time/date, page requested, referral data, and in a separate log browser data), which although now contains no identifiable data, would still in no situation whatsoever be sold or distributed to any third party vendors
This is better than what almost everyone else is doing. Please remember that any search engine proxy (such as StartPage or SearX) has to by definition send the requests to the search engine it is proxying. This means a violator gets at least your query, and probably browser data and the other stuff. Mojeek can avoid this due to using their own index, and keep the data only for themselves. It does store logs though (the ones given above) and the duration of data storage is not stated. Still, finally we have an independent, non-censoring search engine that at least somewhat cares about your privacy. Unfortunately, Mojeek is pretty useless for non-English results.
Ecosia UPDATE June 2021: now Cloudflared, and therefore useless. Their claim to fame has been planting trees for every 45 searches (that's an average) you make using their service - but this, of course, relies on Bing ads being displayed (and I doubt it
can counter the rampant deforestation, anyway). It's not very private by default - For example, when you do a search on Ecosia we
forward the following information to our partner, Bing: IP address, user agent string, search term, and some settings like your country and language setting.
However, it claims to respect
the DNT header: If you have "Do Not Track" enabled in your browser settings we do not collect any analytics data. Most other websites ignore this setting - we think users should have a
choice.
Is the IP address apart of analytics data? Depending on how literally the DNT information is interpreted, Ecosia can become a pretty good choice, combining ethics and privacy (okay, this clearly isn't relevant now after Ecosia became Cuckflared). You
can search for regular sites, Images, Videos, News and Maps. However, the search results are sourced exclusively from Bing. Will work without JavaScript but the images will not be displayed at
all.
Oscobo UPDATE September 2022: can almost never get this to search anymore. An allegedly private one with a not-so-good privacy policy (last updated 2 years ago...). It proudly proclaims several times how it doesn't store information in a way that can identify you
(can you be any more vague), however the myth that this is a privacy-respecting practice has been busted (archive) (MozArchive). And they don't actually say what they DO store and for how long - an indirect
admission that they do store quite a bit of data indeed. Then comes this gem: Oscobo uses cookies to determine the effectiveness of our own marketing campaigns
. And the funniest thing
is I don't see any cookies being set in my uMatrix (outdated privacy policy, as mentioned before; however, it does show they did that at one point, or at least planned to do so). Then it says:
Oscobo uses proprietary technology to hide you search history from others who may use your device after you search. This may save you from some embarrassing situations
But I see the
search queries in my browsing history, plain as a day - so this claim was a mistake too (and a negative one this time). Oscobo can search for images (flickr only - almost useless), videos and
maps (embedding google directly, yawn); it doesn't require JavaScript but images won't show up then. Anyway, due to the issues mentioned earlier, this search engine does not seem trustworthy at
all, and I don't recommend it. It has nothing over the more widely known ones and for something even funnier, it contains a direct link to some shady "Oscobo Browser" executable file. Haha -
avoid.
Discrete Search UPDATE March 2022: This trash doesn't work anymore in Pale Moon or Librewolf. I have no idea why I even checked it out, but there you go.
To work properly, this one needs all of cookies, JavaScript and XHR enabled. What do you get for that? Let's check out their privacy policy (archive) (MozArchive):
Discrete Search does not track search history in any user identifiable way.
The big lie of so-called private search engines rears its ugly head again. The reply is the same as to Oscobo - "user identifiable" is vague and often includes a lot of information that can reveal you if put together (archive) (MozArchive). Why not just mention what the fuck do you store and let us decide whether we're comfortable with it? Graciously, Discrete Search DOES provide that information later (unlike Oscobo):
Additionally, we store aggregated search data to improve product performance, but never store IP addresses or unique user identifiers in connection with such searches in order to ensure that none of the information collected in connection with your searchactivity is personally identifiable.
So aggregated search data
is logged; the important part comes next though. If you read between the lines, they admit that they do store your IP address as well as
unique user identifiers
, just allegedly unconnected with search queries. And that is what you call private? No thanks! Why rely on someone's dubious "unconnecting" when they could just
not store the data? And Discreet Search does seem to store a lot of it, increasing the risk. Then there's this gem from the TOS (archive) (MozArchive):
You certify that you own all intellectual property rights in Your Content. You hereby grant us, our affiliates, and our partners a worldwide, irrevocable, royalty-free, nonexclusive, sub-licensable, license to use, reproduce, create derivative works of, distribute, publicly perform, publicly display, transfer, transmit, distribute, and publish Your Content and subsequent versions of Your Content for the purposes of (i) pursuing our business interests, (ii) distributing Your Content, either electronically or via other media, to third parties seeking to download or otherwise acquire it, and/or (iii) storing Your Content in a remote database accessible by third parties. This license shall apply to the distribution and the storage of Your Content in any form, medium, or technology now known or later developed.
TL;DR everything you submit or transmit to the engine becomes theirs (not just queries but stuff like the headers your browser sends), and they are going to store it in unspecified third party databases. Ha ha. Discrete Search puts fucking image ads on top of your searches - a practice I've never seen anywhere else. At least they've fulfilled the promise of encrypting your searches locally, unlike Oscobo (yeah I've checked). Still, avoid this trainwreck.
Gibiru Claims to be Protecting Your Privacy Since 2009
on its main page. If you read their privacy policy, it all checks out - Gibiru stores no logs or cookies. Of course, what they've
neglected to mention is that they literally use Google directly for their results (without proxying them, like StartPage does or Scroogle used to). Gibiru will not work
without enabling Google scripts, and so all the data that it allegedly doesn't collect, Google will happily take instead. Unfortunately - as you can see - search engines have their fair share
of frauds too. Avoid!
Gigablast Independent search engine, ran since the year 2000. Gigablast's index is surprisingly good, and it does not appear to censor (same as Mojeek). However, it requires JavaScript to display results. Image search shows only thumbnails alongside text, no "gallery" like everyone else does, so you will need to use something else for images if you want that functionality. No third party requests, good privacy policy, preventing third party sharing of your data and deleting logs regularly
- though that is not elaborated on.
Peekier UPDATE May 2023: now redirects to Kagi Search that's paid and seems completely different, review maybe coming soon.
When I learned of Peekier after rating some less-than-stellar search engines, I was delighted to finally find another one that respects the user. There is no IP storage, no third party requests, no logging the data your browser sends (aside from temporary search queries - same policy as DuckDuckGo). It also has the handy feature of displaying the resulting website without visiting it - showing you the relevant information in an efficient way. And then I see this:
Cloudflare, our caching provider, may use a single session cookie for anti-DDOS measures.
Okay, so while YOU might not store anything, Cloudflare - hiding in the shadows between you and me - will happily take it all. And this claim - then - becomes a lie:
SSL/TLS is enforced throughout the website. No unencrypted information is transmitted over the internet.
Because Cloudflare decrypts the request on their servers, it cannot be said that the information is encrypted for its whole journey. Since Cloudflare might be the biggest current Internet evil (archive) (MozArchive), I cannot just let this seemingly small point
slide. Of course, Peekier also claims it does not use cookies in any way to track you or store personally identifiable data
, but it does require a Cloudflare cookie to work (which is
literally a tracking cookie), as well as enabling JavaScript. And that, my friends, ruins a search engine that could have topped the list.
Presearch Some people have inquired about this, so I guess I have to write about it. I'm so tired of these crypto scams, which is the entire reason to exist for Presearch, it seems. You can earn their special crypto token in there by searching. Though there are some serious requirements (archive) (MozArchive) to satisfy:
For Usage Rewards:
1,000 PRE must be earned in order to claim Usage Rewards
A maximum of 1,000 PRE may be claimed every 90 days
Presearch reserves the right to increase or decrease these thresholds at any time
So, you cannot withdraw your funds until you earn at least 70 USD. The really malicious part is that they can change that amount at any time; no problem to start requiring 10 000 or a trillion PRE to withdraw; you cannot ever feel confident that you will be actually able to do so. Many people on their Twitter (MozArchive) are complaining that they can't withdraw their funds even a year later, so this has all the hallmarks of a crypto scam. Presearch can also just steal the tokens you've earned:
If a Member stakes PRE toward search staking and then is determined by our fraud detection system to have generated invalid search or referral activity, then their staked tokens may be forfeited and not recoverable.
Any advertiser can buy their way (archive) into this system:
To enable advertisers to have their ads displayed under certain keywords that the advertisers ‘stake’ their tokens against
The biggest payer for a given keyword wins and gets their ads through to you:
Advertisers can stake their PRE to a keyword, and whichever advertiser stakes the most tokens will have its ads displayed when a user searches on the term selected.
To participate, they will have to buy at least 70 USD worth of the PRE token, for which Presearch charges 10% markup (archive) and that they can of course steal at any time. This is funny too (as if they couldn't make it anymore obvious what they're doing):
If paying by credit card, a 7 month hold on ability to withdraw tokens to the blockchain may be applied to protect against charge backs
You can also host a node, and if you pay 4000 PRE (archive) (MozArchive) (280 USD, who's going to spend that on this scam?!), will also get rewarded for it:
In order to incentivize node operators who commit to providing consistent, stable and responsive computing resources, we are paying Presearch PRE tokens as rewards for node operators who stake a minimum of 4,000 PRE tokens on their node.
FOSS lovers can go home because the node software is closed (archive) (MozArchive):
This software is currently not open source and you are relying on our assurances that nothing malicious is happening underneath the hood. We promise that this is true, but the ability to verify it for yourself would currently be limited to packet sniffers and any other means that you have available to you. It is our intention to open source this software in the future, but we’re not there yet.
Either way, you will almost certainly not earn (archive) (MozArchive) what is required to cover your spendings:
Though the amount paid is variable, we are also ensuring that any node operator who successfully served at least one query receives a minimum of 100 PRE, with numbers increase from there up to about 70x that amount for the most active node operators.
Signing up for an account requires solving malicious hCaptcha, which I obviously didn't bother to do so I don't know what information is required to give otherwise. You don't need to care about this is if you only want to use the engine though - but then, why bother? It's not too good and the results are heavily censored for queries like Moon landing was fake
; Presearch does not have its own index (archive) (MozArchive):
The next step is to begin layering our own index on top of this platform so that community-driven results can supersede the external results
Next step
, meaning it hasn't happened yet. It relies on (archive) (MozArchive) a number of the world’s top search engines and other data sources
, like everyone else. So the claims of it being decentralized
are fake, too. Using Presearch requires JavaScript and cookies to be enabled. Lots of data is collected (archive) (MozArchive):
- Any information you provide through your profile (name, nickname, avatar, etc.)
- Your email address
- Your password (one-way encrypted)
- Any information provided when making a purchase
- Account referrals
- The search engine providers you select
- Search events (when a search was conducted, how many tokens were earned, and which provider was used)
- Your IP Address
- Your blockchain wallet address if you submit or connect it to your account or use it for a deposit or withdrawal.
It is also all shared with shady Business partners
that are not specified:
Partners. We may share transient information with trusted partners during a particular request, for purposes of providing the Service. This may include trusted advertising providers who contractually agree to delete the information after the request is completed so that it is not stored.
When Presearch finally decides to fuck off, they will give their buyer all the data they've collected:
Business Transfers. We may share personal information when we do a business deal, or negotiate a business deal, involving the sale or transfer of all or a part of our business or assets. These deals can include any merger, financing, acquisition, or bankruptcy transaction or proceeding.
As was obvious all along, Presearch is a crypto scam pretending to be a search engine. Its existence is based entirely on advertisers buying their way in (with a 10% cut for Presearch) to shove ads down your throat, and then Presearch doing everything to prevent the exchange of their tokens to a cryptocurrency that is actually useful. But of course people got lured in by promises of easy money and revolutionizing searching through blockchain technology
. Trash it and don't look back.
Right Dao Another search engine with its own index; if you're hopeful, don't be - it is bad. I can't find anything related to conspiracies with my usual queries, though I don't necessarily think they are censoring things - their index is just that weak. Even obscure video game stuff doesn't appear - but does in eg Mojeek. After doing a few more tests with other queries, it appears I can find pretty much nothing I want to, so my verdict is that this search engine is useless. Oh, and no image search - just Web and News. Despite Right Dao being useless, let's check out its privacy (archive) (MozArchive) while I'm here:
IF YOU ARE NOT AT LEAST 16 YEARS OLD, DO NOT USE THE SERVICES.
Hey, I just happen to be 14. What you gonna do, report me to Joe Biden? Moving on...
We do not track our users, and do not collect user identifiable information.
I'm tired of these generic slogans, just tell us what exactly you collect - then we will decide whether we consider that tracking or user identifiable
.
We collect very limited data to improve our service.
Improving search quality: we analyze user submitted queries and result clicks.
Sounds like tracking to me. And not very limited
.
And we do not sell and share such data to any third-parties.
Thanks. At least there is that. But...
We retain the information we collect for as long as necessary to fulfill the purpose and for the purpose of satisfying any legal, accounting, or reporting requirements.
Worthless. Fucking worthless. Tell us the maximum duration, or you're no better than the biggest violators. Might as well have said nothing since this sentence doesn't provide any actionable information to the user (except to stay away from you). Since they admitted to collecting search queries and clicks, and the duration is unlimited, those might be used against you someday. Their ToS (archive) (MozArchive) contains something funny:
Access the Services by any means other than through the interface that is provided by Right Dao [...is not allowed];
You may not send automated queries of any sort to the Services
Curling chads can go home. Anyway, I'm getting tired already. Can I say something positive about this engine? Well, it doesn't discriminate against TOR or Pale Moon. And results appear pretty fast, though they suck. Nevermind, this is pointless. Just skip it.
4get UPDATE February 2024: 4get now has a custom-made captcha required to solve before you can search; and you get 100 searches per solution. If you block cookies, then you get a new captcha for every search, which makes the engine unusable. There are 4get instances that don't have the captcha, though.
New search engine someone recently mentioned to me. It promises to be simple (archive) (MozArchive):
Provide users with a privacy oriented, extremely lightweight, ad free, free as in freedom (and free beer!) way to search for documents around the internet, with minimal, optional javascript code.
And private:
I don't log IP addresses, user agents, or anything else. The npt tokens are the only thing that are stored (in RAM, mind you), temporarly, encrypted.
Just the way I like it :D. Anyway, let's get down to the details. Right off the bat we have an advantage over SearX in that you don't have to fiddle with engines in the settings before you search - you choose one during searching. This means that results get displayed from only one of the source engines, but that is the way I prefer it. In SearX - if I wanted to see only the Mojeek results - I'd have to go to the settings and turn all the other stuff off, which is annoying. What 4get does is much cleaner.
Another advantage is the availability of Yandex and Marginalia as source search engines, which SearX lacks. More importantly, 4get actually lifts results reliably; you avoid the famous failing search engines from SearX, which is what made me give it up completely. Though, this might only be temporary, before the source search engines start banning 4get. Enjoy it while it lasts (UPDATE July 2024: yeah, that's gone now). Overall, SearX has more search engines to choose from, but many of them are duplicates eg DDG and Qwant, or Startpage and Google - which use the same indexes. Still, 4get lacks a Google-based search engine option currently, which would be nice to have. UPDATE July 2024: exists now, as well as some other additions.
SearX has lots and lots more features...but to be honest, I've never needed any of them aside from the Wayback Machine integration - which is just a convenience feature that can be easily replicated by hand (edit: actually, it does exist here if you click the favicon of a search result, to the left of the text). What I don't like, though, is that - if using DDG or Brave source engine - images, videos, or Wikipedia results display on the right side of the page. I get that it's because the source engines themselves do that - but in a simple engine, I'd rather just remove that and only display the actual results. After all, Videos
and Images
are already categories that you can choose. So they shouldn't be mixed in with the textual results that you wanted. Wikipedia is also a propaganda site, so I hate giving it privileged space. I understand that this might be a bitch to code, though.
So, for me this is the best search engine for now. Aside from that small issue I mentioned above, it ticks all the boxes. 4get also has an onion domain. There is no JS dependency, either. Oh, maybe another issue is that the donation service is behind Cloudflare - but not really relevant to the engine itself. IMO, it should just give crypto addresses, and that's that.
FindX I wasn't supposed to review any more search engines, but this one is doing such a great Mozilla impersonation (pretending to be private with the reality being otherwise) that I just had to do this writeup. To be honest, I have never heard of FindX and likely would not have if a reader didn't mention it to me - but the level of its privacy fakery is so great that it deserves a proud spot as the last entry in this report. Let's check out their About page (archive) first:
Findx is a search engine for users who value privacy. We do not collect information about you when you perform searches, unlike most other search engines. Please read the details in our Privacy Policy. If you worry about companies creating invisible profiles of you based on your searches and internet browsing habits - Findx is for you.
The other stuff in there is not very relevant. Remember the above quotes while we inspect their Privacy policy (archive). It starts with the usual GDPR stuff, but interestingly, it seems to try to squirm out of, or not explain properly, many of those rights you allegedly have. For example:
You can request that we restrict usage of your personal data. We are only required to comply in certain circumstances.
And from Section 9:
We may ask you for additional information to confirm your identity and for security purposes, before disclosing the PII requested to you. We reserve the right to charge a fee where permitted by law, for instance if your request is manifestly unfounded or excessive.
Now compare the above with how Qwant treats GDPR (archive) (MozArchive). But since I don't put much value into these laws at all, let's move on to the more
important sections - namely 3.2 Findx Search
, which will tell us what their engine actually collects.
[...] data is transferred to our search partners who only use it to provide better search results on Findx
So one "partner" wasn't enough, there had to be two - CodeFuel and Microsoft. FindX also shamelessly claims that Microsoft values your privacy
- can you believe it? Anyway, what is
the data being sent?
IP address, user agent string, search term, country and language settings, filter settings for adult content, active search filter settings (e.g. page number information), an optional Bing ID (read more below) and the ID of the organization that should benefit from your search.
So the privacy-respecting FindX shares pretty much everything possible with not one, but two third parties! Do they have an explanation for that? Sure they do - but I'm warning you - this will be one of the most idiotic things you've ever read on the Internet. Better wear a helmet, because the stupidity beam is coming!
"But you still pass on my IP-address to CodeFuel and Bing". Yes, we have to. We protect your privacy to the maximum extent allowed under the agreement we have with them. Other privacy-centric search engines mask part of your IP-address before passing it on to their partners - we are not allowed to do that. It is a matter of trust. You have to trust that the privacy search engines do not pass on your full IP-address to their partners. On Findx, you have to trust that our partners only use it for what thay say they will, which is to provide better results on our site (only) - nothing else. We trust them.
Wow, I've been hit with a hurricane of nonsense and I'm so confused I don't even know where to start with this. They literally admit to being worse than all other search engines - since those
(such as Swisscows or StartPage) actually limit the data sent to the providers they use. FindX just says fuck that and dumps all your info onto Bing, then tells you to
just trust their partners to honor an agreement that is never really specified aside from vague claims of "better search results". Sounds dangerously close to Mozilla's improve your experience
while they abuse all your data. And why would Microsoft submit to some small fry like FindX? It's more likely they just do whatever the fuck they want with your
stuff. What about the other partner, CodeFuel?
When search results are displayed, tracking pixels reveal to our search partner which of the results and ads we received were actually seen by you. Our partner, CodeFuel, solely uses this information for building metrics for us and statistics about the service usage. The information is not sold or shared with advertising companies. It simply provides us with information about earnings and various statistics like how many searches resulted in display of ads, how many unique users performed searches, how many results were clicked on etc.
So, FindX assures us that the data collection of CodeFuel is pretty mild. But their privacy policy (archive) (MozArchive) tells a different story:
Certain usage related information regarding your use and interaction with your device, including the Software & Services and Other Software, such as when and how you Use the Software & Services and Other Software, how you use your internet browser and internet search related applications, your language settings, the web pages you visit, applications you use and the content you see, access and utilize on such web pages and applications; for example, offerings and advertisements that you view, use and access, how you use them and your response to them (ie. clickstream data), how frequently you use them, your search queries and the non-precise location, time and date of your searches
Even though FindX explicitly claims in their FAQ (archive) (MozArchive) that this data is not shared with third parties...
CodeFuel acts as a 'middle-man' between small to medium-sized partners and Microsoft, and passes the data on to them in order to serve search results and ads. CodeFuel uses this information themselves to build metrics and statistics about the service usage. The information is not sold to or shared with anybody
CodeFuel does directly state several times that they do so:
In performing fraud detection and prevention, we use the services of third parties who will receive and access your PII.
We also use your PII to comply with legal and regulatory requirements [...] For this purpose, we may share your information with law enforcement or other competent authorities and any third party [...] We will also share this information with our professional advisors in the scope of this purpose.
We may also share your information with our subsidiaries, affiliated and parent companies pursuant to the legitimate interest for the provision of the Software & Services to you but their use of such information must comply with this Privacy Policy.
So two "partners" turned into who the fuck knows how many. There is way more dirt on FindX - their website also collects a bunch of data, TOS pretty much makes you a slave, etc. But I don't want to spend more time on this shitty search engine than absolutely necessary. Needless to say, it's not at all private and in fact, it's probably better to use Bing directly since FindX sends everything there anyway while also running their own spy operation (not to mention CodeFuel and their partners also laying their dirty hands on your data). This is probably the most dishonest search engine out there, rivaling VFEmail and Hushmail in their shamelessness.
UPDATE September 2022: wow, FindX even manages to only give a few results (does not go beyond page 1), and it seems they are all censored (no conspiracy results). Truly the worst search engine in the world.
This search engine is absolutely awesome. It has found me so many interesting sites that are completely missing from the mainstream search engines. Of course, it's completely useless for technical questions and stuff like that; I tried to use it this way for a while but it just doesn't work. Yet, we have Google and Bing for those things already, right? As the creator says: It’s perhaps not the greatest at finding what you already knew was there. Instead it is designed to help you find some things you didn’t even know you were looking for
. And at this task it performs admirably. The index appears to be a lot bigger than Wiby's, while still keeping most mainstream trash away because of heuristics to prefer text-heavy websites
and an upper limit on heavy modern web design inventions
. It is a breath of fresh air for someone that's mostly been chained to the big engines (and their proxies) for all those years. Unfortunately, Marginalia's heuristics
don't get rid of Wikipedia, StackOverflow, and other stuff I'm tired of seeing. Fortunately, there are options to kick those out - such as using either of the filters Small Web
or Blogosphere
(the Forums
filter might be interesting too in an age where all the commotion is on Discord or Facebook groups). When I go back to a "regular" search engine, it just hurts my eyes to see all the nytimes and snopes crap; I wonder how I was able to endure it all those years. Another thing I'm surprised to have endured are the absolutely terrible visual designs of all modern search engines - and Marginalia resists this disease, too. Look:
It actually uses colors and margins to separate entries, while all the other search engines just throw the entire mess right onto the background. In Marginalia, half the horizontal space of the screen also isn't taken up by a Wikipedia entry (ugh) or just empty space (like in searx, 4get, mojeek, DDG...). It just looks and functions great; it is obvious someone with a good eye for design created it. You are even told which websites contain ads, cookies, or scripts, and can also deny those that do have JS or ads. Though this is not that useful for uMatrix chads, it's still nice to see someone even caring to divide websites by these criteria. I'd like to see Cloudflare as another criteria in there, while we're at it.
About privacy, it's almost as private as it can be. Access logs are only kept for 24h, and After this window is passed, all logs that may contain user queries are shredded
. Also No information about which links are clicked is gathered, and it is not possible to historically correlate IP address to search terms
. So are there any flaws? Again, it doesn't find "normal" things well at all - which is also its advantage and of course we have all the others that do this. Sometimes the results don't have a lot to do with the exact query, so treat this engine as an exploratory device, and not necessarily an informational one. I don't know the creator's exact stance on censorship; he does say on his about page that if you do find something genuinely in bad taste
, you should report it. But what's the criteria, I don't know. I seem to be finding all kinds of "inappropriate" stuff that would surely be banned on mainstream search engines; I suspect it's mostly about porn and violence. There are also no image results or anything other than just sites - but again, we can use other things for those. For once, it's nice to have something that does one thing and one thing well. In the end, Marginalia is one of the best software projects I've seen recently, after it dropped Cloudflare.
No idea what this is, except that it is Cloudflared, and therefore useless.
Same as above.
Same as above.
Same as above.
Same as above.
Same as above.
Same as above.
A search engine that rejects mainstream sites. Could be cool, but is unfortunately Cloudflared.
So, the first edition of this summary was overly positive. I was eager to finally share some good news after the stuff on Mozilla, Proton, and others dirtied the privacy climate. However, I severely overestimated the situation and way overrated several search engines such as StartPage, MetaGer and DuckDuckGo. The main issue is that all of them lack their own index, meaning they are fully dependent on the violators. The ones who do use their own index have weak results. This already proves that the situation is worse than with browsers (which has the independent Pale Moon as well as various other smaller, less viable projects) or E-mail. Not only is there no provider approaching the quality of RiseUp mail - but even the second-tier of E-mail providers (such as Posteo or Dismail) outclass the best search engines. The field is also rife with frauds, similar to E-mail. What's a poor user to do, then? It would be best to pull the plug and deal with the inconvenience of terrible results, though of course it's hard to actually do that. However, the situation won't improve until we support the independent search engines. So, use Mojeek and Wiby to accomplish that. If you're not prepared to ditch the violators, a good SearX instance is your best bet. I recommend weaning yourself off Google since they're so heavily censored, therefore StartPage is out. The wounded privacy warrior marches on, scraping by until a decent provider finally comes along...
| Engine / Feature | Requires JavaScript | Own index | Supports TOR | Has onion domain | IP storage | Search results | Censorship | Shares data with third parties |
| Swisscows | Yes | No | No | No | No | Good | Mild | Yes |
| Brave Search | For images | Doubtful | Sometimes | Yes | No | Average | Extreme | No |
| SearX | No | No | Yes | Conditional | Conditional | Good | None | No |
| StartPage | No | No | No | No | No | Good | Extreme | No |
| Wiby | No | Yes | Yes | No | Yes | Bad | None | No |
| DuckDuckGo | No | Partial | Yes | Yes | Yes | Good | Mild | No |
| Qwant | No | Partial | Yes | No | Yes | Good | Mild | Yes |
| Metager | No | No | Yes | Yes | Yes | Good | None | No |
| Mojeek | No | Yes | Yes | No | No | Average | None | No |
| Ecosia | For images | No | No (CF) | No | Yes | Good | None | Yes |
| Oscobo | For images | No | Yes | No | Unknown | Good | Mild | Yes |
| Discrete Search | Yes | No | Yes | No | Yes | Good | None | Yes |
| Gibiru | Yes | No | No | No | No | Good | Extreme | Yes |
| Gigablast | Yes | Yes | Yes | No | Yes | Average | None | No |
| Peekier | Yes | Partial | No (CF) | No | Yes | Good | None | Yes |
| Presearch | Yes | No | Yes | No | Yes | Good | Extreme | Yes |
| Right Dao | No | Yes | Yes | No | Possible | Bad | Who knows | No |
| 4get | No | No | Yes | Yes | No | Good | None | No |
| FindX | No | No | Yes | No | Possible | Bad | Extreme | Yes |
| Marginalia | No | Yes | Yes | No | Yes | Average | Little / who knows | No |
| You | Cloudflare so who cares lol | |||||||
| WebCrawler | ||||||||
| Search my Site | ||||||||
| MillionShort | ||||||||
| Spyda | ||||||||
| Excite | ||||||||
| 4Conservative | ||||||||
| GoodGopher | ||||||||
Explanation of issues:
Conditionalbecause only a few instances have that.
Conditional. Also remember to contrast this category with
Supports TOR- if a service is red in both, it's really bad.
Nonebecause it doesn't censor by itself and you can choose the non-censorious engines in it. But if you rely on the censorious ones, of course SearX then inherits the grade of those.
You might be wondering as to why I didn't review the Cloudflared engines properly. That is because using Cloudflare is such a big flaw that the other qualities of the search engine don't even matter. Cloudflare is a third party that MitMs your connection and steals your search query and all other sent data, blocks TOR, and is also able to rewrite the served page without you knowing. I can't see anything possibly worse than that.
UPDATE July 2024: I've been questioning the way these criteria are set up recently, I don't find them too descriptive. IP storage is treated as one thing, when it might be for one day or forever. Query storage is missed completely. Censorship has many meanings and might refer to censorship of alternative theories, porn, violence, or just JS-heavy sites. So, the table remains to be improved but for now, I hope it can still provide some useful information