Decentralization is Killing Our Environment
Decentralization is becoming prevalent in our networked society. As we become advanced, interconnected beings, we should consider the potentially adverse effects that could result in a cumulative negative experience.
The “Fediverse” refers to users communicating and sharing over the network, using open-source software hosted on many servers. It typically refers to social networking, virtual show-and-tell, however it could include machines and devices which share information with people or other machines or devices. If the server software is programmed according to the published W3 “Activity Pub” standards, it is interoperable with other server software that is also programmed according to the standards. The interconnectivity is considered as “federation”. The servers are federated, like a country of unique, individual states. Conceptually, one could imagine a world where users of Facebook, Twitter, and TikTok could all communicate with each other seamlessly, sharing photos of their kids, pets, vacations, and whatnot. In real life, these social networking corporations do not play nicely with each other. A user may “share” a video stored on one platform, on another platform, however the two platforms are not integrated.
Anyone with gumption can set up a VPS and install the open-source federated server software of their choice. The server software “Mastodon” is the most popular. According to Fediverse Observer, there are approximately five million total users in the Fediverse, and 3.5 million (70%) use Mastodon. Note that a user is not compelled to run a node on the network, they can instead opt to join an established node which openly accepts registrations from the public. Mastodon users can communicate and connect to the other 30% of the users in the Fediverse.
A user on the federated network can connect to other users as “friending”, which subscribes the users to other’s posts. When a user looks at their message feed on the website, they see their posts, comments, and replies, as well as the posts of their connections (and other users who may not have a connection, but have commented or favored a post in the thread.)
The Activity Pub standard is designed to be “decentralized”. This means that all the posts and comments are not stored in one database, they are distributed throughout the Fediverse. All the federated servers make HTTP requests to each other, which happens behind the scenes and is transparent to most average users.
The software uses two methods of PKI (public key infrastructure) to sign and authenticate every message, similar to signing a cryptocurrency transaction on a blockchain, such as Bitcoin. The message is signed, and as a second operation, the request headers.
Each user has a private and a public key. Only the private key can sign messages, the public key used to verify the signature. Users of the federated software may not even realize they have a private and public encryption key because the server manages everything in the background without their knowledge or involvement.
Signing and verifying messages is computationally expensive, requiring increased CPU utilization.
The Anatomy Of The Message
I will now examine a single message sent to me by a Mastodon user. The message is a sentence that is 135 characters, including spaces. Inspecting the HTTP request headers, we can see the message size increased from 135 characters to 13,937 characters. (over 10,000% increase). This is because the message is overloaded with metadata “bloat”. The metadata describes the source of the message, other messages it references, information about the type of message, etc. This message contains no encoded content such as images, audio, or video. A simple “favorite” or “like” will become a message of about 2,500 characters.
Since the system is “decentralized”, this message is copied to all subscribers. If the poster has 100 “friends” then 100 copies go out, similar to an email with 100 recipients. However, the message continues to be replicated to subscribers of the subscribers. This can turn into thousands of duplicate messages spread out across the Fediverse. To determine the “Eco-footprint” of a single message, there’s network traffic, CPU processing in signatures and verifications, and data storage of all these messages.
Consider someone sitting in front of their computer “liking” everything all day, creating a massive Eco-footprint.
I am running a test on a simple website at dekameta.com. I sent a connection request to my account I have on another federated server, then sent a single message to that account. A few of my connections on that server made a comment about the post. I decided to send a connection request to any account that sends a message to my inbox. Without scanning or scraping any federated server directories, I have so far made over 12,000 connection requests and over 8,000 have accepted and followed me back. I have received over a million messages in my inbox since I started the experiment about a month ago. It’s mind-boggling that a single message (to myself) generated that much traffic. The stats are updated in real time at https://dekameta.com/about
