Gopher (allegedly from "go for information") is a network protocol for publishing, browsing and downloading files and is known as a much simpler alternative to the World Wide Web (i.e. to HTTP and HTML). In fact it competed with the Web in its early days and even though the Web won in the mainstream, gopher still remains used by small communities (usually the more dedicated though, see e.g. bitreich). Gopher is like the Web but well designed, it is the suckless/KISS way of doing what the Web does, it contains practically no bloat and so we highly advocate its use. Gopher inspired creation of Gemini, a similar but bit more complex and "modern" protocol, and the two together have recently become the main part of so called Smol Internet. Gopher is much better than Gemini though. The set of all public gopher servers is called gopherspace. The Gopher protocol was defined in 1993 in RFC 1436.
Gopher doesn't use any encryption (though some servers allow access via Tor). This is good, encryption is bloat. Gopher also doesn't really know or care about Unicode and similar bloat (which mostly serves trannies to insert emojis of pregnant men into readmes anyway, we don't need that), it's basically just ASCII (of course you can employ Unicode as gopher just transfers files really, it's just that Unicode is not part of gopher's specification and most people prefer to keep it ASCII). Gopher's simple design is intentional, the authors deemed simplicity a good feature. Gopher is so simple that you may very well write your own client and server and comfortably use them -- you can even browse gopher just by manually using telnet to communicate with the server.
How big is/was gopherspace? In 1994 there were over 1300 gopher servers (source: 1994 book Finding it on the Internet"), around 1995 there were already more than 6000 (source: 1995 video Searching the Internet - Gopher | The Internet Revealed). Most of them are now gone, in 2005 there were only 145 servers reported by Veronica search engine (source: 2007 gopher archive), though Gopher recently saw a new wave of popularity. As of 2023 the Veronica search engine reported 315 gopher servers in the world with 5+ million indexed selectors, which they estimated was 83% of the whole gopherspace (the peak server count was in 2020 at almost 400). Quarry search engine reports 369 servers and 1+ million indexed selectors. Contrition search engine reported even 495 servers and 7+ million selectors. The "grawler" crawler of gopherspace.de reported 192 active servers at the beginning of 2016, 182 in 2020, 413 in 2023 (bumped its search list) and 380 in 2024. Gopher LAWN directory (made by bitreich) contains 281 selected quality gopher holes.
From the user's perspective the most important distinction from the Web is that gopher is based on menus instead of "webpages"; a menu is simply a column of items of different predefined types, most importantly e.g. a text file (which clients can directly display), directory (link to another menu), text label (just shows some text), binary file etc. A menu can't be formatted or visually changed, there are no colors, images, scripts or hypertext -- a menu is not a presentation tool, it is simply a navigation node towards files users are searching for (but the mentioned ASCII art and label items allow for somewhat mimicking "websites" anyway). Gopher is also often browsed from the command line, though graphical clients are a thing too. Addressing works with URLs just as the Web, the URLs just differ by the protocol part (gopher:// instead of http://), e.g.: gopher://gopher.floodgap.com:70/1/gstats. What on Web is called a "website" on gopher we call a gopherhole or just hole (i.e. a collection of resources usually under a single domain) and the whole gopher network is called a gopherspace. Blogs are common on gopher and are called phlogs (collectively a phlogosphere). As menus can refer to one another, gopher creates something akin a global file system, so browsing gopher is like browsing folders and can comfortably be handled with just 4 arrow keys. Note that as menus can link to any other menu freely, the structure of the "file system" is not a tree but rather a general graph. Another difference from the Web is gopher's great emphasis on plaintext and ASCII art as it cannot embed images and other media in the menus (even though of course the menus can link to them). There is also a support for sending text to a server so it is possible to implement search engines, guest books, games etc.
Gopher is just an application layer protocol (officially running on port 70 assigned by IANA), i.e it sits above lower layer protocols like TCP and takes the same role as HTTP on the Web and so only defines how clients and servers talk to each other -- the gopher protocol doesn't say how menus are written or stored on servers. Nevertheless for the creation of menus so called gophermaps have been established, which is a simple format for writing menus and are the gopher equivalent of Web's HTML files (just much simpler, basically just menu items on separate lines, the exact syntax is ultimately defined by server implementation). A server doesn't have to use gophermaps, it may be e.g. configured to create menus automatically from directories and files stored on the server, however gophermaps allow users to write custom menus manually. Typically in someone's gopherhole you'll be served a welcoming intro menu similar to a personal webpage that's been written as a gophermap, which may then link to directories storing personal files or other hand written menus. Some gopher servers also allow creating dynamic content with scripts called moles.
Gopher software: sadly "modern" browsers are so modern they have millions of lines of code but can't be bothered to support such a trivial protocol like gopher, however there are Web proxies you can use to explore gopherspace (look up e.g. floodgap). Better browsers such as lynx (terminal), sacc, clic or forg (GUI) can be used for browsing gopherspace natively (it's not hard, you don't need to learn any keybinds, using arrow keys usually just works). As a server you may use e.g. Gophernicus (used by SDF) or search for another one, there are dozens. { Personally I've used gophrier for server, it was the simplest one I found. ~drummyfish } For the creation of gophermaps you simply use a plaintext editor. Where to host gopher? Pubnixes such as SDF, tilde.town and Circumlunar community offer gopher hosting but many people simply self-host servers e.g. on Raspberry Pis, it's pretty simple.
A quick tl;dr/sumup of the gopher world/community as of 2023: thankfully there doesn't seem to be much censorship and/or woke toxicity that's seen on the web, the community is still quite small, which is probably a GOOD thing, though something like doubling the gopherspace size would probably still be welcome; an issue/downside at this time seems to be a "self serving" nature of gopher ("come to gopher to learn about gopher"), i.e. instead of gopher being a "platform" for sharing all kinds of information, we mostly have a gopher community talking about gopher, so outsiders really have nothing to come in for. Of course there is interesting information of other kinds, but the overall impression is just this. Another issue is that current search engines like Veronica don't seem to be fulltext but rather search only document titles. Currently gopherspace seems to be mostly divided into following gopherholes:
Some basic/interesting gopher links: gopher://bitreich.org/1/lawn (directory of gopher holes), gopher://gopher.floodgap.com/7/v2/vs (search engine), gopher://circumlunar.space:70/1 (circumlunar space), gopher://gopherpedia.com (Wikipedia on gopher), gopher://gopher.icu/7/quarry (search engine), ...
Here is a nice tutorial: git clone git://bitreich.org/gopher-tutorials/.
To quickly try browsing gopher either use a web proxy, e.g. at https://gopher.floodgap.com/gopher/gw.lite, or use some nice native browser, e.g. lynx gopher://floodgap.com.
More technical details: just as with the web, you have some gopher server running somewhere (some IP address/domain, on port 70) which serves resources to clients. A client connects to the server (via TCP) and simply sends the name of the resource (file or directory) it wants to retrieve as a string ending with a newline. If the string is empty, the server sends the default directory (the "main page"). You may try this manually in terminal using telnet, nc or a similar tool. For example:
echo "" | nc floodgap.com 70
The server just sends us back a list of available resources in the "main directory", one per line, each in format:
<TYPE><DISPLAY_STRING><TAB><PATH><TAB><SERVER><TAB><PORT>
For example one of the lines here looks like this:
1Search Gopherspace with Veronica-2 <TAB> /v2 <TAB> gopher.floodgap.com <TAB> 70
Here 1 says the resource is a directory, then we have a display string (which you'll see in the browser), then the path to the resource, i.e. /v2, and then the server and port. If we want to retrieve this directory, we send:
echo "/v2" | nc floodgap.com 70
And get a similar response. This is basically all a client needs to know.
As for running a server, details depend on each one, but generally they behave like this: you have a server running in some default directory, let's say /home/me/my_gopherhole. By default a server will just serve list of files present in this directory to clients who request the "main directory", treating directories as subdirectories and sending regular files back. However there is one important feature: you may create a gophermap file to create a custom menu, or something akin a "gopher website". Gophermap is something like gopher's HTML, just much more simple. How to do this? You simply create a file name gophermap in the directory (the main one or any subdirectory) -- if the server sees such a file, it serves it instead of listing the directory file.
TODO: continue
TODO
Powered by nothing. All content available under CC0 1.0 (public domain). Send comments and corrections to drummyfish at disroot dot org.