Created in February of 2007, Tumblr is something of the dark horse of social media. It’s a notoriously unprofitable platform that routinely struggles to expand its user base. Despite this, Tumblr’s established audience (most of whom have been active on the site for more than twelve years) are fervent defenders of both the site’s format and culture.

There are many things about Tumblr that make it stand apart: for a long time its most notable feature was its lack of an algorithm. When a new user boots up the site they’re prompted to choose a minimum of five tags to follow from a pre-generated list of categories. The list is extensive and each category is its own tag with sub-categories clearly listed. After following the tags the user is led to follow three blogs (other users) that regularly post about one of the chosen tags. This is because a user’s Tumblr feed is not influenced by their likes or watchtime; it’s purely based on what they choose to follow.

Nowadays Tumblr has implemented a “based on your likes” algorithm in an attempt to conform to standard web-trends. Like many of the site’s newer features, the option to disable it manually was added after unilateral backlash from the users. And for good reason! A chronological feed exclusively made up of posts from people you follow or topics you care about is unheard of in today’s social media landscape. The absence of an algorithm also removes the potential for “endless” scrolling - go past whatever is newly posted and you start hitting things you’ve already seen in chronological order. It makes for a convenient solution to the common issue of social media addiction, you don’t need to worry about missing anything - you simply run out of things to see.

Getting “popular” on Tumblr is both laughably easy and infinitely more difficult than on other sites; it’s all about who you know. In the absence of an algorithm, posts gain traction through “reblogs”. A “reblog” is a form of interaction that essentially reuploads a post to your own blog. it retains all the post’s data but gets displayed to any followers you have in the same feed as any of your own original posts. This means a post made on Tumblr can snowball very easily: you make a post; a few of your followers reblog it - then a few of their followers reblog it and so on and so forth. This means a small blog with only ten or so followers could regularly be making semi-viral posts if one of those followers happens to have a large audience themselves.

Conveniently, on the desktop version of the site every post has a built-in graph of reblog chains. Each “dot” represents a reblog and the lines connecting them chart which thread caused that blog to find the post. Clicking on each one of these dots leads to the blog it relates to.

Tumblr has lots of useful stat trackers and external sites like jetblackcode fill in the blanks of what the main platform doesn’t provide. On-site you can track your blog’s daily interaction and filter notifications by likes; reblogs, comments etc. Off-site, jetblackcode.com can total the stats of original posts on a blog; calculate your average notes per post, list the blogs you most reblog from etc. The total notes on a post is a combination of likes; reblogs and replies. A single “note” is one instance of any one of those three things. The number of notes is listed at the bottom of any given post, clicking on the notes opens a post’s stats page where you can read comments and view its reblog chain graph/likes.

Reblogs as a mechanic of social media are unique in they essentially become a new post. The OP (original poster) still receives notifications from interaction and any notes gained from a reblog are combined into the post’s overall stats. However, if a post is deleted, the reblogs remain - functionally, all deleting a post on Tumblr does is produce an error message when the origin of the post is clicked on. In order to fully “delete” a post, each individual instance of that post being reblogged must also be deleted. To avoid this, users can turn off reblogs in a post’s settings, a feature only implemented in 2022.

The back-end of Tumblr’s interface is unknown but speculation suggests it has a similar issue. Occasionally there will be incidents of “ghost” posts - posts stuck at 0 notes because they’ve been removed from Tumblr server-side and thus the space to store their data no longer exists. These are rare - likely created by staff mistakes and by nature difficult to search for without a link. The most well known of these is a text post that reads “How is bnha anime of the decade...... they aren’t even anime of the hour of the minute of the second”. If it was recorded properly, the true notes on this post would be upward of a million. As it stands it appears to have no reblogs or comments and a small number of likes. The likes are unique to the incidence of the post created by the reblog; likes in general are stored twice - once in total on the Tumblr servers and once tracking the new likes generated by each reblog. Therefore, the reblogs aren't tracked but the likes are; albeit individually to each copy of the post created by the reblog. This is proven by the fact that reblogging the same “ghost” post to a different blog resets the notes.

This quirk of reblogging combined with a search engine that struggles with older posts has generated a subculture of themed archival blogs like @heritageposts, a blog dedicated to archiving the more infamous posts on the site. It’s an interesting read, everything archived uses a wildly outdated fandom slang that would be considered universally “cringe” in today’s era. It’s a snapshot of a more naive internet - a time when being strange online was the norm. Many early Tumblr posts were characterised by an unquestioning belief in obvious misinformation. The efforts of @heritageposts and blogs like it may seem silly (and what they archive often is), but the posts they go through the effort of finding and cataloging are more often than not created by now-deleted blogs and would never be easily accessible otherwise. 

As mortifying as it is to be unable to completely take down a post if it starts gaining traction for the wrong reasons, the alternative may be worse. Take for example Twitter, or “X” as it's now called. In 2023 Elon Musk essentially destroyed any potential for unbiased archival work on the platform. By locking Twitter’s API (Application Programing Interface) behind an obscene a paywall, 3rd party programers can no longer access and document a tweet based on its metadata. Twitter is a far more politically relevant platform than Tumblr (the last major politician to have a publicly known account on Tumblr was Barack Obama @obamawhitehouse, but that blog hasn’t been updated since January 2017), and yet a deleted Twitter post means a burned paper trail. There’s no method of verifiably saving a tweet beyond screenshots, which are easily faked. Tumblr by comparison allows the user to cut ties with a post: turning off reblogs and deleting it severs any connection it has with the original poster in regards to notifications and search results, but the post still exists. This makes it difficult to falsify statements from a user, as not being able to provide a copy of the original post is fairly damning evidence for it not having existed in the first place.

In general, the site’s original design philosophy built itself around anonymity and customisation. Tumblr supports changes to a blog’s html code and allows users to create almost entirely unique themes for their page. This, in combination with the ability to create sideblogs (blogs attached to a primary account that can have multiple moderators and even change ownership between accounts) lends itself to an entirely personalised experience.

The focus on anonymity has remained less intact. In 2023, Tumblr changed to an access model requiring an account in order to browse the site and send anonymous messages. This was credited as an attempt to mitigate harassment and according to Tumblr itself was very effective. It’s hard not to see it as part of a trend though. Increasing efforts have been made by staff to conform to the current state of the internet; either to attract new people or to extract a profit from current users – sometimes both. The site has been criticised for the now-backpedaled implementation of a livestreaming feature as well as various methods for users to monetize their content: all of which were removed after backlash (X, X).

The simple fact is, the Tumblr user base is actively hostile to new core features designed for profit. It's not that they’re allergic to change; polls were a relevantly recent addition (2023) and were received with great enthusiasm. The site has cultivated a culture with little to no interest being advertised to, or in the case of its creators – leveraging their audience for the purpose of a career. It’s a deeply unprofessional platform built to foster small communities around a particular interest. The lack of algorithm; minimal data collection and (comparatively) small-scale audience mean large advertisers have little interest in the site.

So, what remains for Tumblr? It may feel counterintuitive but users don’t want paid services to add to the function of the site. Locking content behind a paywall or relying on creators to be compelled to profit from their contributions to the site is unsustainable in a space so reliant on collaboration. 

When Twitter (“X”) changed their verification system to a paid model, Tumblr openly mocked the rival social media by offering paid “verification” of their own. For the low price of $7.99 (one cent below the twitter blue’s price of $8) you could buy two blue checkmarks next to your name that did absolutely nothing. As of writing, this is still a feature you can buy today. The checkmarks even stack!

This open defiance of functionality both resonated with users and effectively advertised the site. Mocking Elon’s decisions cemented the site as a sort of anti-Twitter; something a lot of people were actively searching for at the time. According to CEO Matt Mullenweg, in the week following Twitter’s acquisition, Tumblr’s app store downloads increased by more than 50% across the board. The characterisation of the site as a janky but down to earth bastion of the old internet was something that had existed before the controversy, but the contrast with Twitter brought it into the mainstream. It’s an image cultivated intentionally by Tumblr staff. According to their own research, Gen Z and Millennials resonate best with “meaningful engagement” and “community” – they want the brands to relate to them.

And it works! The checkmarks reportedly increased Tumblr’s revenue by 125%. Users who can pay often do so out of a desire to maintain the site, not for any particular service. The adblocker subscription is convenient but also a service that can be applied to the entire internet for free by anyone with a passing knowledge of computers. The checkmarks and option to “send crabs” (pay $3 to fill another user’s dashboard with small, moving crabs images for 24 hours) are fundamentally useless and not trying to change the core experience of the site. This reputation and the level of connection cultivated with the brand has resulted in users organising to give Tumblr money as if it were a charity event.

Tumblr is ranked 32 globally as a social media. It continues to lose money despite the promising research it’s made into its core user base and unless something big changes that deficit is likely to continue. Still, there’s something important about the site; in today’s internet there’s nothing quite like it. While a company like any other, the service it provides – an entirely customizable social media experience potentially unaltered by the inputs of an algorithm – is unique and should be protected.