-- Leo's gemini proxy
-- Connecting to envs.net:1965...
-- Connected
-- Sending request
-- Meta line: 20 text/gemini
Originally posted 2021-03-10. Last updated 2024-05-01.
This is a cursory review of all the indexing search engines I have been able to find. Gemini engines are at the bottom; the rest of this post is about Web search engines.
The three dominant English search engines with their own indexes¹ are Google, Bing, and Yandex (GBY). Many alternatives to GBY exist, but almost none of them have their own results; instead, they just source their results from GBY.
With that in mind, I decided to test and catalog all the different indexing search engines I could find. I prioritized breadth over depth, and encourage readers to try the engines out themselves if they’d like more information.
This page is a “living document” that I plan on updating indefinitely. Check for updates once in a while if you find this page interesting. Feel free to send me suggestions, updates, and corrections; I’d especially appreciate help from those who speak languages besides English and can evaluate a non-English indexing search engine. Contact info is in the article footer.
I plan on updating the engines in the top two categories with more info comparing the structured/linked data the engines leverage (RDFa vocabularies, microdata, microformats, JSON-LD, etc.) to help authors determine which formats to use.
I discuss my motivation for making this page in the "Rationale" section.
I primarily evaluated English-speaking search engines because that’s my primary language. With some difficulty, I could probably evaluate a Spanish one; however, I wasn't able to find many Spanish-language engines powered by their own crawlers.
I mention details like "allows site submissions" and structured data support where I can only to inform authors about their options, not as points in engines' favor.
See the "Methodology" section at the bottom to learn how I evaluated each one.
These are large engines that pass all my standard tests and more.
1. Google: the biggest index. Allows submitting pages and sitemaps for crawling, and even supports WebSub to automate the process. Powers a few other engines:
*formerly* Startpage; it switched to Bing. Line the results up against DuckDuckGo or Ecosia to see for yourself.
GMX search
Mullvad Leta (leta.mullvad.com)
(discontinued) Runnaroo
SAPO (Portuguese interface, can work with English results)
DSearch
13TABS
Zarebin.ir (Persian, can return English results)
A host of other engines using Programmable Search Engine's client-side scripts.
2. Bing: the runner-up. Allows submitting pages and sitemaps for crawling without login using the IndexNow API. Its index powers many other engines:
Yahoo (and its sibling engine, OneSearch)
DuckDuckGo³
Startpage
AOL
Qwant (partial)⁴
Ecosia
Ekoru
Privado
Findx
Disconnect Search⁵
PrivacyWall
Lilo
SearchScene
Peekier
Oscobo
Million Short
Yippy search⁶
Lycos
Givero
Swisscows
Fireball
Netzzappen
You.com¹¹
Partially powers MetaGer by default; this can be turned off
At this point, I mostly stopped adding Bing-based search engines. There are just too many.
3. Yandex: originally a Russian search engine, it now has an English version. Some Russian results bleed into its English site. Like Bing, it allows submitting pages and sitemaps for crawling using the IndexNow API. Powers:
Epic Search (went paid-only by June 2021)
Occasionally powers DuckDuckGo’s link results instead of Bing. (update: DuckDuckGo has "paused" its partnership with Yandex)
Petal for Russian users only.
4. Mojeek: Seems privacy-oriented with a large index containing billions of pages. Quality isn’t at Google/Bing/Yandex’s level, but it’s not bad either. If I had to use Mojeek as my default general search engine, I’d live. Partially powers eTools.ch. At this moment, I think that Mojeek is the best alternative to GBY for general web search.
Google, Bing, and Yandex support structured data such as microformats1, microdata, RDFa, Open Graph markup, and JSON-LD. Yandex's support for microformats1 is limited; for instance, it can parse h-card metadata for organizations but not people. Open Graph and Schema.org are the only supported vocabularies I'm aware of. Mojeek is evaluating structured data; it's interested in Open Graph and Schema.org vocabularies.
These engines pass most of the tests listed in the "methodology" section. All of them seem relatively privacy-friendly. I wouldn't recommend using these engines to find specific answers; they're better for learning about a topic by finding interesting pages related to a set of keywords.
Stract: My favorite generalist engine on this page. Stract supports advanced ranking customization by allowing users ti import "optics" files, like a better version of Brave's "goggles" feature. Stract is fully open-source, with code released under an AGPL-3.0 license. The index is isn't massive but it's big enough to be a useful supplement to more major engines. Stract started with the Common Crawl index, but now uses its own crawler. Plans to add contextual ads and a subscription option for ad-free search. Discovered in my access logs.
Right Dao: very fast, good results. Passes the tests fairly well. It plans on including query-based ads if/when its userbase grows.⁸ For the past few months, its index seems to have focused more on large, established sites rather than smaller, independent ones. It seems to be a bit lacking in more recent pages.
Alexandria: A pretty new "non-profit, ad free" engine, with freely-licensed code. Surprisingly good at finding recent pages. Its index is built from the Common Crawl; it isn't as big as Gigablast or Right Dao but its ranking is great.
Yep: an ambitious engine from Ahrefs, an SEO/backlink-finder company, that "shares ad profit with creators and protects your privacy". Most engines show results that include keywords from or related to the query; Yep also shows results linked by pages containing the query. In other words, not all results contain relevant keywords. This makes it excellent for less precise searches and discovery of "related sites", especially with its index of *hundreds of billions of pages.* It's far worse at finding very specific information or recent events for now, but it will probably improve. It was known as "FairSearch" before its official launch.
SeSe Engine: Although it's a Chinese engine, its index seems to have a large-enough proportion of English content to fit here. The engine is open-source. It has surprisingly good results for such a low-budget project. Each result is annotated with detailed ranking metadata such as keyword relevance and backlink weight. Discovered in my access logs.
greppr: its tagline is "Search the Internet with no filters, no tracking, no ads." At the time of writing, it has over 3 million pages indexed. It's surprisingly good at finding interesting new results for broad short-tail queries, if you're willing to scroll far enough down the page. It appears to be good at finding recent pages.
Yep supports Open Graph and some JSON-LD at the moment. A look through the source code for Alexandria and Gigablast didn't seem to reveal the use of any structured data. The surprising quality of results from SeSe and Right Dao seems influenced by the crawlers' high-quality starting locations (e.g. Wikipedia).
These engines fail badly at a few important tests. Otherwise, they seem to work well enough.
Infotiger: My favorite engine in this section. It offers advanced result filtering and sports a somewhat large index. It allows site submission for English and German pages. The fastest-improving engine in this section; I look forward to the day it "graduates" to the previous section. Infotiger also has a Tor hidden service.
seekport: The interface is in German but it supports searching in English just fine. The default language is selected by your locale. It’s really good considering its small index; it hasn’t heard of less common terms. but it’s able to find relevant results in other tests. It's the second-fastest-improving engines in this section.
Exalead: slow, quality is hit-and-miss. Its indexer claims to crawl the DMOZ directory, which has since shut down and been replaced by the Curlie directory. No relevant results for “Oppenheimer” and some other history-related queries. Allows submitting individual URLs for indexing, but requires solving a Google reCAPTCHA and entering an email address.
ExactSeek: small index, disproportionately dominated by big sites. Failed multiple tests. Allows submitting individual URLs for crawling, but requires entering an email address and receiving a newsletter. Webmaster tools seem to heavily push for paid SEO options. It also powers SitesOnDisplay and Blog-search.com.
Burf.co: Very small index, but seems fine at ranking more relevant results higher. Allows site submission without any extra steps.
Siik: Lacks contact info, and the ToS and Privacy Policy links are dead. Seems to have PHP errors in the backend for some of its instant-answer widgets. If you scroll past all that, it does have web results powered by what seems to be its own index. These results do tend to be somewhat relevant, but the index seems too small for more specific queries.
ChatNoir: An experimental engine by researchers that uses the Common Crawl index. The engine is open source. There's more information in its announcement on the Common Crawl mailing list (Google Groups).
Secret Search Engine Labs: Very small index with very little SEO spam; it toes the line between a "search engine" and a "surf engine". It's best for reading about broad topics that would otherwise be dominated by SEO spam, thanks to its CashRank algorithm. Allows site submission.
Gabanza: a search engine from a hosting company. I found few details abou the search engine itself, and the index was small, but it was suitable for discovering new pages related to short broad queries.
Results from these search engines don’t seem at all useful.
Yessle: seems new; allows page submission by pasting a page into the search box. Index is really small but it crawls new sites quickly. Claims to be private.
Bloopish: extremely quick to update its index; site submissions show up in seconds. Unfortunately, its index only contains a few thousand documents (under 100 thousand at the time of writing). It's growing fast: if you search for a term, it'll start crawling related pages and grow its index.
YaCy: community-made index; slow. Results are awful/irrelevant, but can be useful for intranet or custom search.
Scopia: only seems to be available via the MetaGer metasearch engine after turning off Bing and news results. Tiny index, very low-quality.
Artado Search: Primarily Turkish, but it also seems to support English results. Like Plumb, it uses client-side JS to fetch results from existing engines (Google, Bing, Yahoo, Petal, and others); like MetaGer, it has an option to use its own independent index. Results from its index are almost always empty. Very simple queries ("twitter", "wikipedia", "reddit") give some answers. Supports site submission and crowdsourced instant answers.
Active Search Results: very poor quality
Crawlson: young, slow. In this category because its index has a cap of 10 URLs per domain. I initially discovered Crawlson in the seirdy.one access logs. This is often down; if the current downtime persists, I'll add it to the graveyard.
Anoox: Results are few and irrelevant; fails to find any results for basic terms. Allows site submission. It's also a lightweight social network and claims to be powered by its users, letting members vote on listings to alter rankings.
Yioop!: A FLOSS search engine that boasts a very impressive feature-set: it can parse sitemaps, feeds, and a variety of markup formats; it can import pre-curated data in forms such as access logs, Usenet posts, and WARC archives; it also supports feed-based news search. Despite the impressive feature set, Yioop's results are few and irrelevant due to its small index. It allows submitting sites for crawling. Like Meorca, Yioop has social features such as blogs, wikis, and a chat bot API.
Spyda: A small open-source engine made by James Mills, written in Go.
Slzii.com: A new web portal with a search engine. Has a tiny index dominated by SEO spam. Discovered in the seirdy.one access logs.
Engines in this category fall back to GBY when their own indexes don't have enough results. As their own indexes grow, some claim that this should happen less often.
Brave Search: Many tests (including all the tests I listed in the "Methodology" section) resulted results identical to Google, revealed by a side-by-side comparison with Google, Startpage, and a Searx instance with only Google enabled. Brave claims that this is due to how Cliqz (the discontinued engine acquired by Brave) used query logs to build its page models and was optimized to match Google.¹⁰ The index is independent, but optimizing against Google resulted in too much similarity for the real benefit of an independent index to show. Furthermore, many queries have Bing results mixed in; users can click an "info" button to see the percentage of results that came from its own index. The independent percentage is typically quite high (often close to 100%) but can drop for advanced queries. (update: Brave's Bing contract appears to have expired as of April 2023).
I can't in good conscience recommend using Brave Search, as the company runs cryptocurrency, has held payments to creators without disclosing that creators couldn't receive rewards, has made dangerously misleading claims about fingerprinting resistance (will update with a link to my thoughts on the matter), is run by a CEO who spent thousands of dollars opposing gay marriage, and has rewritten typed URLs with affiliate links.
Plumb: Almost all queries return no results; when this happens, it falls back to Google. It's fairly transparent about the fallback process, but I'm concerned about *how* it does this: it loads Google's Custom Search scripts from "cse.google.com" onto the page to do a client-side Google search. This can be mitigated by using a browser addon to block "cse.google.com" from loading any scripts. Plumb claims that this is a temporary measure while its index grows, and they're planning on getting rid of this. Allows submitting URLs, but requires solving an hCaptcha. This engine is very new; hopefully as it improves, it could graduate from this section. Its Chief Product Officer previously founded the Gibiru search engine which shares the same affiliates and (for now) the same index; the indexes will diverge with time.
Qwant: Qwant claims to use its own index, but it still relies on Bing for most results. It seems to be in a position similar to Neeva. Try a side-by-side comparison to see if or how it compares with Bing.
Kagi Search: Like Neeva, it requires an account and limits use without payment. It's powered by its own Teclis index (Teclis can be used independently; see the non-commercial section below), and claims to also use results from Google and Brave. The result seems somewhat unique: I'm able to recognize some results from the Teclis index mixed in with the mainstream ones. In addition to Teclis, Kagi's other products include the Kagi.ai intelligent answer service and the TinyGem social bookmarking service, both of which play a role in Kagi.com in the present or future. Unrelatedly: I'm concerned about the company's biases, as it seems happy to use Brave's commercial API (allowing blatant homophobia in the comments) and allow its results to recommend suicide methods without intervention. I reject the idea that avoiding an option that may seem politically biased is the same as being unbiased if such a decision has real political implications.
SVMetaSearch: A SearxNG metasearch engine that also includes results from its own index. All other sources can be turned off. Like most public Searx/SearxNG instances, reliability is very poor.
These indexing search engines don’t have a Google-like “ask me anything” endgame; they’re trying to do something different. You aren't supposed to use these engines the same way you use GBY.
Marginalia Search: A recent addition similar to Wiby, and *my favorite entry on this page*. It has its own crawler but is strongly biased towards non-commercial, personal, and/or minimal sites. It's a great response to the increasingly SEO-spam-filled SERPs of GBY. Partially powers Teclis, which in turn partially powers Kagi. Update 2022-05-27: Marginalia.nu is now open source
Ichido: An engine that just rolled out its own independent index, with a lot of careful thought put into its ranking algorithm. Like Marginalia, it's biased towards the non-commercial Web: it downranks ads, CAPTCHAs, trackers, SEO, and obfuscation.
Teclis: A project by the creator of Kagi search. Uses its own crawler that measures content blocked by uBlock Origin, and extracts content with the open-source article scrapers Trafilatura and Readability.js. This is quite an interesting approach: tracking blocked elements discourages tracking and advertising; using Trafilatura and Readability.js encourages the use of semantic HTML and Semantic Web standards such as microformats, microdata, and RDFa. It claims to also use some results from Marginalia. The Web interface has been shut down, but its standalone API is still available for Kagi customers.
Clew: a FOSS new engine with a small index of several thousand pages. It focuses on independent content and downranks ads and trackers; there seems to be a real focus on quality over quantity, which makes it excellent for short-tail searches (especially around technical concepts). Ranking is more egalitarian than other engines, making it better for discovery and surfing than research. It's designed to be small and lightweight, with a compact index. Discovered in the seirdy.one access logs.
These engines try to find a website, typically at the domain-name level. They don't focus on capturing particular pages within websites.
Kozmonavt: The best in this category. Has a small but growing index of over 8 million sites. If I want to find the website for a certain project, Kozmonavt works well (provided its index has crawled said website). It works poorly for learning things and finding general information. I cannot recommend it for anything serious since it lacks contact information, a privacy policy, or any other information about the org/people who made it. Discovered in the seirdy.one access logs.
search.tl: Generalist search for one TLD at a time (defaults to .com). I'm not sure why you'd want to always limit your searches to a single TLD, but now you can.⁹ There isn't any visible UI for changing the TLD for available results; you need to add/change the "tld" URL paramater. For example, to search .org sites, append "&tld=org" to the URL. It seems to be connected to Amidalla.de. Amidalla allows users to manually add URLs to its index and directory; I have yet to see if doing so impacts search.tl results.
Thunderstone: A combined website catalog and search engine that focuses on categorization. Its about page claims: "We continuously survey all primary COM, NET, and ORG web-servers and distill their contents to produce this database. This is an index of *sites* not pages. It is very good at finding companies and organizations by purpose, product, subject matter, or location. If you’re trying to finding things like 'BillyBob's personal beer can page on AOL', try Yahoo or Dogpile." This seems to be the polar opposite of the engines in the "small or non-commercial Web" category.
sengine.info: Developed by netEstate GmbH, which specializes in content extraction for inprints and job ads. Also has a German-only version available. Discovered in my access logs.
Gnomit: Allows single-keyword queries and returns sites that seem to cover a related topic. I actually kind of enjoy using it; results are old (typically from 2009) and a bit random, but make for a nice way to discover something new. For instance, searching for "IRC" helped me discover new IRC networks I'd never heard of.
High Browse: uses a non-traditional ranking algorithm which does an excellent job of introducing non-SEO-optimized serendipity into search results. One of my favorite "surf-engines", as opposed to traditional "search-engines".
Keybot: A must-have for anyone who does translation work. It crawls the web looking for multilingual websites. Translators who are unsure about how to translate a given word or phrase can see its usage in two given languages, to learn from other human translators. My parents are fluent English speakers but sometimes struggle to express a given Hindi idiom in English; something like this could be useful to them, since machine translation isn't nuanced enough for every situation. Part of the TTN Translation Network. Discovered in my access logs.
Quor: seems to mainly index large news sites. Site is down as of June 2021. Originally available at www dot quor dot com.
Semantic Scholar: a search engine by the Allen Institute for AI focused on academic PDFs, with a couple hundred million papers indexed. Discovered in my access logs.
Bonzamate: a search engine specifically for Australian websites.
searchcode: A code-search engine by the developer of Bonzamate. Searches a hand-picked list of code forges for source code, supporting many search operators.
Lixia Labs Search: A new engine that focuses on indexing technical websites and blogs, with a minimal JavaScript-free front-end. Discovered in my access logs. Surprisingly good results for broad technical keyword queries.
I’m unable to evaluate these engines properly since I don’t speak the necessary languages. English searches on these are a hit-or-miss. I might have made a few mistakes in this category.
Baidu: Chinese. Very large index; it's a major engine alongside GBY. Offers webmaster tools for site submission.
Qihoo 360: Chinese. I’m not sure how independent this one is.
Toutiao: Chinese. Not sure how independent this one is either. Its index appears limited outside of its own content distribution platform.
Sogou: Chinese
Yisou: Chinese, by Yahoo. Appears defunct.
Naver: Korean. Allows submitting sitemaps and feeds. Discovered via some Searx metasearch instances.
Daum: Korean. Also unsure about its independence.
Seznam: Czech, seems relatively privacy-friendly. Discovered in the seirdy.one access logs. It allows site submission with webmaster tools and using IndexNow.
Cốc Cốc: Vietnamese
go.mail.ru: Russian
ALibw.com: Chinese, found in my access logs.
Vuhuv: Turkish
search.ch: Regional search engine for Switzerland; users can restrict searches to their local regions.
fastbot: German
Moose.at: German (Austria-based)
SOLOFIELD: Japanese
kaz.kz: Kazakh and Russian, with a focus on "Kazakhstan's segment of the Internet"
These engines come close enough to passing my inclusion criteria that I felt I had to mention them. They all display original organic results that you can't find on other engines, and maintain their own indexes. Unfortunately, they don't quite pass.
Wiby: I love this one. It focuses on smaller independent sites that capture the spirit of the “early” web. It’s more focused on “discovering” new interesting pages that match a set of keywords than finding a specific resources. I like to think of Wiby as an engine for surfing, not searching. Runnaroo occasionally featured a hit from Wiby (Runnaroo has since shut down). If you have a small site or blog that isn’t very “commercial”, consider submitting it to the index. Does not qualify because it seems to be powered only by user-submitted sites; it doesn't try to "crawl the Web".
Mwmbl: like YaCy, it's an open-source engine whose crawling is community-driven. Users can install a Firefox addon to crawl pages in its backlog. Unfortunately, it doesn't qualify because it only crawls pages linked by hand-picked sites (e.g. Wikipedia, GitHub, domains that rank well on Hacker News). The crawl-depth is "1", so it doesn't crawl the whole Web (yet).
Search My Site: Similar to Marginalia and Teclis, but only indexes user-submitted personal and independent sites. It optionally supports IndieAuth. Its API powers this site's search results; try it out using the search bar at the bottom of this page. Does not qualify because it's limited to user-submitted and/or hand-picked sites.
Blog Surf: a search engine for blogs with RSS/Atom feeds. Does not qualify because all blogs submitted to the index require manual review, but it seems interesting. Its "MarketRank" algorithm seems to give it a bias towards sites popular on "Hacker" "News".
Kukei.eu: a curated search engine for web developers, which crawls a hand-picked list of sites. As it does not index the whole Web, it doesn't qualify. I still find it interesting.
Ask.com: The site is back. They claim to outsource search results. The results seem similar to Google, Bing, and Yandex; however, I can’t pinpoint exactly where their results are coming from. Also, several sites from the "ask.com network" such as directhit.com, info.com, and kensaq.com have uniqe-looking results.
Not evaluated: Apple’s search. It’s only accessible through a search widget in iOS and macOS and shows very few results. This might change; see the next section.
Not evaluated: Kagi Search. It's in a closed beta and I haven't yet gotten an invitation.
Partially evaluated: Infinity Search. It has a young, small index. It recently split into a paid offering with the main index and Infinity Decentralized, the latter of which allows users to select from community-hosted crawlers. I managed to try it out before it became a paid offering, and it seemed decent; however, I wasn’t able to run the tests listed in the “Methodology” section. Allows submitting URLs and sitemaps into a text box, no other work required.
Some search engines are integrated into other appliances, but don’t have a web portal.
Apple’s search engine is usable in the form of “Siri Suggested Websites”. Its index is built from the Applebot web crawler. If Apple already has working search engine, it’s not much of a stretch to say that they’ll make a web interface for it someday.
Amazon bought Alexa Internet (a web traffic analysis company, at the time unrelated to the Amazon Alexa virtual assistant) and discontinued its website ranking product. Amazon still runs the relevant crawlers, and also have a bot called “Amazonbot”. While Applebot powers the Siri personal assistant, Amazonbot powers the Alexa personal assistant “to answer even more questions for customers”. Crawling the web to answer questions is the basis of a search engine.
Time for my first Gemini-exclusive content! A Gemini page about search engines wouldn't be complete without a few search engines for the Gemini space.
geminispace.info: A GUS instance, but with an updated index. Supports submitting content. The biggest search engine on Gemini.
AuraGem Search Engine: part of the Ponix capsule, written in Go. A relative newcomer.
These engines were originally included in the article, but have since been discontinued.
Petal search: A search engine by Huawei that recently switched from searching for Android apps to general search. Despite its surprisingly good results, I wouldn't recommend it due to privacy concerns: its privacy policy describes advanced fingerprinting metrics, and it doesn't work without JavaScript. Requires an account to submit sites. I discovered this via my access logs. Be aware that in some jurisdictions, it doesn't use its own index: in Russia and some EU regions it uses Yandex and Qwant, respectively. Inaccessible as of June 2023.
Neeva: formerly in the "semi-independent" section. Combined Bing results with results from its own index. Bing normally isn't okay with this, but Neeva was one of few exceptions. Results were mostly identical to Bing, but original links not found by Bing frequently popped up. Long-tail and esoteric queries were less likely to feature original results. Required signing up with an email address or OAuth to use, and offered a paid tier with additional benefits.
Gigablast: It’s been around for a while and also sports a classic web directory. Searches are a bit slow, and it charges to submit sites for crawling. It powered Private.sh. Gigablast was tied with Right Dao for quality. Shut down mid-2023.
gus.guru: the original Gemini search engine. The index doesn't seem to be updated anymore.
wbsrch: In addition to its generalist search, it also had many other utilities related to domain name statistics. Failed multiple tests. Its index was a bit dated; it had an old backlog of sites it hadn’t finished indexing. It also had several dedicated per-language indexes.
Gowiki: Very young, small index, but showed promise. I discovered this in the seirdy.one access logs. It was only available in the US. Seems down as of early 2022.
Meorca: A UK-based search engine that claims not to "index pornography or illegal content websites". It also features an optional social network ("blog"). Discovered in the seirdy.one access logs.
Ninfex: a "people-powered" search engine that combines aspects of link aggregators and search. It lets users vote on submissions and it also displays links to forums about submissions.
Marlo: Another FLOSS engine, written in Haskell. Has a small index that's good enough for surfing broad topics, but not good enough for specific research.
websearchengine.org and tuxdex.com: Both were run by the same people, powered by their inetdex.com index. Searches are fast, but crawls are a bit shallow. Claims to have an index of 10 million domains, and not to use cookies. The pages are currently down and the domains re-direct to porn sites; I'm not aware of any official notice.
Entfer: a newcomer that let registered users upvote/downvote search results to customize ranking. Didn't offer much information about who made it. Its index was small, but it did seem to return results related to the query.
Parsijoo: Persian search engine
Two engines were excluded from this list for having a far-right focus.
One engine was excluded because it seems to be built using cryptocurrency in a way I'd rather not support.
Some fascinating little engines seem like hobbyist proofs-of-concept. I decided not to include them in this list, but watch them with interest to see if they can become something viable.
Why bother using non-mainstream search engines?
Google, Microsoft (the company behind Bing), and Yandex aren't just search engine companies; they're content and ad companies as well. For example, Google hosts video content on YouTube and Microsoft hosts social media content on LinkedIn. This gives these companies a powerful incentive to prioritize their own content. They are able to do so even if they claim that they treat their own content the same as any other: since they have complete access to their search engines' inner workings, they can tailor their content pages to better fit their algorithms and tailor their algorithms to work well on their own content. They can also index their own content without limitations but throttle indexing for other crawlers.²
One way to avoid this conflict of interest is to *use search engines that aren't linked to major content providers;* i.e., use engines with their own independent indexes.
There's also a practical, less-ideological reason to try other engines: different providers have different results. Websites that are hard to find on one search engine might be easy to find on another, so using more indexes and ranking algorithms results in access to more content.
No search engine is truly unbiased. Most engines' ranking algorithms incorporate a method similar to PageRank, which biases them towards sites with many backlinks.
Search engines have to deal with unwanted results occupying the confusing overlap between SEO spam, shock content, and duplicate content. When this content’s manipulation of ranking algos causes it to rank high, engines have to address it through manual action or algorithm refinement. Choosing to address it through either option, or choosing to leave it there for popular queries after receiving user reports, reflects bias. The best solution is to mix different ranking algorithms and indexes instead of using one engine for everything.
I find new engines by:
Monitoring certain web directories for changes in their search engine listings.
Checking other curated lists of "good/bad bots" to spot search engines.
Using search engines to discover search engines: searching for the names of less-popular engines often pulls up similar lists.
Receiving suggestions from readers
Compiling a list of regular expressions for user-agent strings I'm familiar with. Before I delete my server access logs, I extract user-agents that don't match that list along with the pages they request.
Checking the Searx and Searxng projects for new integrations.
Engines in this list should have their own indexes powered by web crawlers. Original results should not be limited to a set of websites hand-picked by the engine creators; indexes should be built from sites from across the Web. An engine should discover new interesting places around the Web.
Here's an oversimplified example to illustrate what I'm looking for: imagine somone self-hosts their own personal or interest-specific website and happens to get some recognition. Could they get *automatically* discovered by your crawler, indexed, and included in the first page of results for a certain query?
I'm willing to make two exceptions:
1. Engines in the "semi-independent" section may mix results that do meet the aforementioned criteria with results that do not.
2. Engines in the "almost qualified" section may use indexes primarily made of user-submitted or hand-picked sites, rather than focusing primarily on sites discovered organically through crawling.
The reason the second exception exists is that while user submissions don't represent automatic crawling, they do at least inform the engine of new interesting websites that it had not previously discovered; these websites can then be shown to other users. That's fundamentally what an alternative web index needs to achieve.
I'm not usually willing to budge on my "no hand-picked websites" rule. Hand-picked sites will be ignored, whether your engine fetches content through their APIs or crawls and scrapes their content. It's fine to use hand-picked websites as starting points for your crawler (Wikipedia is a popular option).
I only consider search engines that focus on link results for webpages. Image search engines are out of scope, though I *might* consider some other engines for non-generalist search (e.g., Semantic Scholar finds PDFs rather than webpages).
I focused almost entirely on "organic results" (the classic link results), and didn't focus too much on (often glaring) privacy issues, "enhanced" or "instant" results (e.g. Wikipedia sidebars, related searches, Stack Exchange answers), or other elements.
I compared results for esoteric queries side-by-side; if the first 20 results were (nearly) identical to another engine’s results (though perhaps in a slightly different order), they were likely sourced externally and not from an independent index.
I tried to pick queries that should have a good number of results and show variance between search engines. An incomplete selection of queries I tested:
“vim”, “emacs”, “neovim”, and “nvimrc”: Search engines with relevant results for “nvimrc” typically have a big index. Finding relevant results for the text editors “vim” and “emacs” instead of other topics that share the name is a challenging task.
“vim cleaner”: should return results related to a line of cleaning products rather than the Correct Text Editor.
“Seirdy”: My site is relatively low-traffic, but my nickname is pretty unique and visible on several of the highest-traffic sites out there.
“Project London”: a small movie made with volunteers and FLOSS without much advertising. If links related to the movie show up, the engine’s really good.
“oppenheimer”: a name that could refer to many things. Without context, it should refer to the physicist who worked on the atomic bomb in Los Alamos. Other historical queries: “magna carta” (intermediate), “the prince” (very hard).
Some less-mainstream engines have noticed this article, which is great! I've had excellent discussions with people who work on several of these engines. Unfortunately, this article's visibility also incentivizes some engines to optimize specifically for any methodology I describe. I've addressed this by keeping a long list of test queries to myself. The simple queries above are a decent starting point for simple quick evaluations, but I also test for common search operators, keyword length, and types of domain-specific jargon. I also use queries designed to pull up specific pages with varying levels of popularity and recency to gauge the size, scope, and growth of an index.
Professional critics often work anonymously because personalization can damage the integrity of their reviews. For similar reasons, I attempt to try each engine anonymously at least once by using a VPN and/or my standard anonymous setup: an amnesiac Whonix VM with the Tor Browser. I also often test using a fresh profile when travelling, or via a Searx instance if it supports a given engine. When avoiding personalization, I use "varied" queries that I don't repeat verbatim across search engines; this reduces the likelihood of identifying me. I also attempt to spread these tests out over time so admins won't notice an unusual uptick in unpredictable and esoteric searches. This might seem overkill, but I already regularly employ similar methods for a variety of different scenarios.
I didn't try to avoid personalization when testing engines that require account creation. Entries in the "hit-and-miss" and "unusable" sections got less attention: for instance, I didn't spend a lot of effort tracking results over time to see how new entries got added to them.
I avoided "natural language" queries like questions, focusing instead on keyword searches and search operators. I also mostly ignored infoboxes (also known as "instant answers").
What I learned by building this list has profoundly changed how I surf.
Using one engine for everything ignores the fact that different engines have different strengths. For example: while Google is focused on being an "answer engine", other engines are better than Google at discovering new websites related to a broad topic. Fortunately, browsers like Chromium and Firefox make it easy to add many search engine shortcuts for easy switching.
When talking to search engine founders, I found that the biggest obstacle to growing an index is getting blocked by sites. Cloudflare was one of the worst offenders, although it has since launched a "verified bots" program to improve things:
Too many sites block perfectly well-behaved crawlers, only allowing major players like Googlebot, BingBot, and TwitterBot; this cements the current duopoly over English search and is harmful to the health of the Web as a whole.
Too many people optimize sites specifically for Google without considering the long-term consequences of their actions. One of many examples is how Google's JavaScript support rendered the practice of testing a website without JavaScript or images "obsolete": almost no non-GBY engines on this list are JavaScript-aware.
When building webpages, authors need to consider the barriers to entry for a new search engine. The best engines we can build today shouldn't replace Google. They should try to be different. We want to see the Web that Google won't show us, and search engine diversity is an important step in that direction.
Try a "bad" engine from lower in the list. It might show you utter crap. But every garbage heap has an undiscovered treasure. I'm sure that some hidden gems you'll find will be worth your while. Let's add some serendipity to the SEO-filled Web.
Some of this content came from the Search Engine Map and Search Engine Party. A few web directories also proved useful.
Matt from Gigablast also gave me some helpful information about GBY which I included in the "Rationale" section. He's written more about big tech in the Gigablast blog:
Nicholas A. Ferrell of The New Leaf Journal wrote a great post on alternative search engines.
He also gave me some useful details about Seznam, Naver, Baidu, and Goo:
¹ Yes, “indexes” is an acceptable plural form of the word “index”. The word “indices” sounds weird to me outside a math class.
² Matt from Gigablast told me that indexing YouTube or LinkedIn will get you blocked if you aren't Google or Microsoft. I imagine that you could do so by getting special permission if you're a megacorporation.
³ DuckDuckGo has a crawler called DuckDuckBot. This crawler doesn't impact the linked results displayed; it just grabs favicons and scrapes data for a few instant answers. DuckDuckGo's help pages claim that the engine uses over 400 sources; my interpretation is that at least 398 sources don't impact organic results. I don't think DuckDuckGo is transparent enough about the fact that their organic results are proxied. Compare DuckDuckGo side-by-side with Bing and you'll see it's sourcing organic results from one of them (probably Bing). Update 2022: DuckDuckGo has the ability to downrank results on its own; it was previously working with Bing to get Bing to remove misinformation and spam:
⁴ Qwant claims to also use its own crawler for results, but it’s still mostly Bing in my experience. See the "semi-independent" section.
⁵ Disconnect Search allows users to have results proxied from Bing or Yahoo, but Yahoo sources its results from Bing.
⁶ Yippy claims to be powered by a certain IBM brand (a brand that could correspond to any number of products) and annotates results with the phrase “Yippy Index”, but a side-by-side comparison with Bing and other Bing-based engines revealed results to be nearly identical.
⁸ This is based on a statement Right Dao made in on Reddit:
⁹ Some search engines support the "site:" search operator to limit searches to subpages/subdomains of a single site or TLD. "site:.one", for instance, limits searches to websites with the ".one" TLD.
¹⁰ More information can be found in a HN subthread and the Cliqz tech blog:
¹¹ I'm in the process of re-evaluating You.com. It claims to operate a crawler and index. As of right now, it seems very much like DuckDuckGo to me: organic results look like they're from Bing, while infoboxes ("apps") seem to be scraped or queried from hand-picked websites; I'm not currently seeing results from "around the web" like the other engines that do pass my inclusion criteria. I might be wrong! I'm re-evaluating it to see if this isn't actually the case. (Update: You.com seems to source organic link fresults from Bing, and only interleaves those results with its own curated infoboxes)
---
Copyright © 2023 Rohan “Seirdy” Kumar
-- Response ended
-- Page fetched on Fri May 3 11:32:22 2024