Getting started with SAMIZDAT
Once you get Samizdat up and running, it is time to get to know how Samizdat works!
Gateway to the network
Simply put, Samizdat Node works as a proxy between your browser (Firefox, Chrome, Safari, etc…) and the Samizdat Network. It spawns a local site in your own computer. According to the Rules of the Internet©, this site is called http://localhost:4510.
Ok… it is not your run-off-the-mill site name, such as
https://www.google.com or
https://web.whatsapp.com,
but it is as valid a name. localhost is a special name that means “the site
inthis computer” (since it is where Samizdat node is running, after all) and
4510 is the number of a non-standard TCP port. I hope you have gotten
the joke.
Let’s try it then! In case you didn’t notice, this very page is hosted in the Samizdat Network. If you can acess it without installing Samizdat Node, chances are that you are accessing it via an external proxy, such as https://proxy.hubfederation.com. This is not the fun way to use Samizdat (and it can be a vulnerability). If you have Samizdat Node installed, you can access this very same page by going to this page:
Got here? To this exact page? Congratulations! You were able to fetch this Samizdat documentation page from the Samizdat Network itself. How meta is that!?
So… what is actually going on here?
Content addressing: what’s in a name?
If you pay close attention to any regular old URL in the regular Web, you will
see that it is comprised of at least two parts: a server and a route. For
example, in https://mail.google.com/mail,
mail.google.com is the server and /mail is the route. This means that there
is a physical computer in some Google warehouse somewhere that answers to you
as mail.google.com. It its the name of this computer. The route, /mail
is just the name of a resource (a file) in that computer. Everything a (valid)
URL is saying is: “there is a file in a computer”. This is called location
addressing, because you reach the content via its location.
SAMIZDAT works with something called content addressing, which is fundamentally
different from location addressing. In content addressing, URLs point to the
content directly, regardless of location. For each file in the network, a
unique name is created that is based solely on the content of the file. This
is called a hash. For example,
this is a link of a picture whose hash is -u3S2coR1vjKwLh6mJ8LOHMY3vnwgCrCsrmt_g:
http://localhost:4510/_objects/-u3S2coR1vjKwLh6mJ8LOHMY3vnwgCrCsrmt_gw
This name is intrinsic to this picture and is guaranteed to be unique.
Now, the key idea behind content addressing over location addressing is that whereas the file https://mail.google.com/mail has to be in Google’s computer and Google has complete control over it, the above link is not bound to any particular device and therefore no one controls it. In fact, many nodes in the network may have copies of the file and any of these nodes can serve it to you. Even better, now that you have the file in your local node (if you have clicked the link above), you are automatically serving it to others.
Don’t worry, though! The underlying protocol makes sure that only those who know the hash (because they have come about a link) can know the contents of the file. Besides, all transactions are carried out anonymously (actually, pseudonymously) so that your privacy is preserved. Of course, SAMIZDAT does not preclude the use of proxies, VPNs or, better still, things like Tor.
Ergonomics
Even though content addressing using hashes is great, it is impossible to have a comfortable web experience using only hashes. There are two reasons for this:
- Hashes are just unintuitive blobs of letters; it is not possible to structure content systematically. More simply put, hashes are ugly.
- You cannot have cyclic links! That is, if page A has a link to page B, page B cannot have a link to page A. Imagine not beign able to link back from your blog post to your homepage!
- This is not easy to see at first, but has to do with the cryptographic guarantees of hash functions.
To solve this, SAMIZDAT has the concept of collections. Collections work just like folders in your computer: you can put your files into folders and subfolders so that your content is organized in trees. Incidentally, URL routes also work the same way. Collections work using these nice data structures called Merkle trees. These trees allow us to verify that a given hash corresponds to the right path inside a collection. After all, beign duped in the Internet can have harsh consequences. As an added bonus, each collection is uniquely identified by a single hash, just like objects, no matter the size. To accessing a collection item from your browser is very similar to accessing an object. For example, that picture I showed you can also be accessed by the link
http://localhost:4510/_collections/pKxHqTFD_Oezb3q7WKKJz6u2_NTwEdKSwAJH0w/example-object.png
Now, the picture has a name: example-object.png, inside collection pKxHqTFD_Oezb3q7WKKJz6u2_NTwEdKSwAJH0w.
However, you will rarely see links to collections because collections are immutable! Once they are built, they cannot be changed, not even a single bit. If you change the bit, the whole collection becomes another collection, with a completely and unpredictably different hash. Therefore, SAMIZDAT has also the concept of series.
A SAMIZDAT series is a sequence of collections, which can be advanced with a new element at any time. It works similarly to editions of a book: once the book is printed, it cannot be changed. The ink has dried and the orders have shipped. However, the author can always create a second, third and fourth version of the book, while removing old editions from print and completely disavowing them.
A series is just a cryptographically signed document containing a collection together with a timestamp, which is its publication date. There may be many signed collections floating around the SAMIZDAT network, but the SAMIZDAT node will try to use the most recent one it finds to give you the freshest content available. Please note that this is a best-effort process and that you are not guaranteed to end up with the latest-and-greatest. You are also not guaranteed that once you publish a new edition that the old ones will disappear from the network. It is easy to put content in the SAMIZDAT network, but it is quite a task to erase seomthing from it.
The page you accessing now is probably beign accessed using a series. If not, here is the link again as a refresher:
Note that the prefx now is _series instead of _collections. Now, the big blob of letters is a public key associated with the secret key used to sign new editions. This is (by now) the preferred way of retrieving content from the SAMIZDAT network and the one which will feel more familiar to most users.
You might be wondering if it is possible to remove that last blob of letters from the URL and substitute it for something small and memorable. It certainly is, but it depends on something called global consensus, something which is more in the realm of blockchains and cryptocurrencies. Don’t fear: a solution is in the workings. However, big fat letters blobs will have to do, for now.
So, that’s it! Happy surfing.
TL; DR
Just go surf te Web as you normally do and SAMIZDAT will take care of the rest for you! But remember: things inside the network are referenced by what they are (content addressing) and not by where they are (location addressing). Even that is not a big of a deal, since the original Web experience was already created to give you this illusion.