Algorithms Manipulate You. Make Your Own Instead.

Algorithms Manipulate You. Make Your Own Instead

Source: Algorithms Manipulate You. Make Your Own Instead., Odysseas, 20:06, uploaded 2026-06-25, Watch Later position 48.

Odysseas begins with Larry Ellison, Oracle’s founder and, according to the video, one of the richest people in the world. Ellison and his son David own Paramount Skydance. The video recounts Paramount’s offers for Warner Bros., Netflix’s attempt to buy the company, and the Ellisons’ hostile takeover, which was still awaiting regulatory approval when the video was published. Their holdings would include Warner Bros., CNN, Adult Swim, Discovery Channel and TLC.

The business story becomes a way to talk about power. Oracle held a 15 percent stake in the arrangement that transferred TikTok from China to the United States, the video says, which gives the company an unusual position around the service’s algorithm. Odysseas also mentions Ellison’s political donations and relationships, including a reported $26 million donation to the IDF and his friendship with Donald Trump. The details are there to establish that a small number of wealthy people can sit close to the systems that decide what millions of people see. The larger claim follows quickly: algorithms create culture, so the people who control them have cultural power.

Algorithms as instruments of power

Odysseas frames the argument as a question of interests rather than a simple hatred of rich people. A small group controls the systems, and that group brings its own ideologies, aims and business needs into the design. Reed Hastings, Netflix’s former CEO, gives the bluntest example. The company’s greatest competitor, Hastings says, is sleep. A recommendation system therefore competes for a part of the day that belongs to the user before it competes with another media company.

The financial incentive is easy to state. Global social-media advertising spending is projected to reach $320 billion in 2026, according to the figure quoted in the video. Every extra view can create another opportunity to show an advert, so the system has a reason to keep attention moving. Odysseas admits the tension in his own position because advertising also supports his work. His objection concerns the scale and direction of the pressure. Social-media algorithms become tools that sell people to advertisers, shape political influence and extract as much attention as possible.

That pressure does not require a secret plan. A business can pursue revenue through ordinary market logic and still produce a system that changes how people think. Odysseas’s practical conclusion is that an individual can reclaim some agency by choosing what to consume instead of allowing the feed to choose the next item.

The five stages of a recommendation feed

Before discussing resistance, Odysseas explains what he means by an algorithm. The word reaches back to Muhammad ibn Musa al-Khwarizmi, the ninth-century Persian polymath whose writing on calculation was translated into Latin in the twelfth century. His name became associated with computation through algorismus, which later became “algorithm”. In its modern sense, the word describes a set of rules for calculation or another problem-solving process.

The social-media version is a recommendation algorithm. Netflix suggests a show, YouTube supplies another video essay, and Instagram can place a disaster or a violent news clip in front of someone who has only just woken up. The platform can show one item at a time, so its business depends on finding material that keeps a person inside the app and creates advertising revenue. Odysseas describes a generalised five-stage process:

The platform catalogues its content and the interactions around it. The collection can include hundreds of millions of videos, billions of tweets and the clicks, views and other actions attached to them.
An automatic moderation filter removes material that is illegal, vile or against the platform’s terms. Other posts remain available yet receive less distribution when the system marks them as offensive or problematic. Odysseas uses shadow banning as the name for this grey area, whilst admitting that the question of what counts as offensive is itself unresolved.
An approximate nearest-neighbour search selects a manageable sample from the full catalogue. The video gives roughly 100 items as an example because ranking everything would take too much time and computing power.
A deep-learning model ranks those candidates with information about the user and the content. The two questions it tries to answer are whether the person will click and whether they will keep watching or engaging.
The platform re-ranks the feed as a whole. The individual predictions have to produce a feed that feels fresh, diverse and balanced enough to make sense to the person using it.

The explanation is deliberately general. Platforms use different systems, and the video gives no technical documentation for the five stages or for the estimate of 100 candidates. Its point concerns the objective that joins the stages together. The system has to predict what will hold attention.

Infinite attention and the need for replacement

The older internet, in Odysseas’s account, gave a person a finite amount of material to choose from. An album ended. A person caught up with the new uploads and reached the bottom of a feed. The activity could stop because the collection had an edge.

Recommendation feeds remove that edge. The content can be well-made and useful, yet the feed still has a commercial reason to continue. More views create more advertising opportunities and more revenue. Odysseas calls addiction the goal and compares the political response to the regulation of tobacco. He mentions European rules that address notifications, autoplay and infinite scrolling through a person’s right to remain undisturbed.

The proposed response starts with replacement. Scrolling creates a positive feedback loop in which frequent stimulation makes ordinary activity harder to begin. Quitting without preparing another use for the time can leave a vacuum that sends the person back to the feed. Odysseas suggests moving gradually towards books, documentaries, newspapers, articles, skills, hobbies, exercise, work and social life. He keeps the standard modest. A person does not have to choose between spending an entire evening on TikTok and working on a grand project for twenty hours. A healthier middle ground can contain ordinary activities with some depth.

He then gives eight ways to reduce the pull of the feed:

Keep the phone out of the bedroom, especially at night and in the morning.
Keep it out of sight and reach whilst working.
Turn off as many notifications as possible and use a minimalist phone or app.
Set the app timers and limits built into modern phones.
Uninstall the applications that cause the problem when they have no good reason to remain.
Use browser extensions such as Social Focus to remove Shorts feeds and recommendations.
Give sustained attention to personal interests and keep them meaningful.
Practise boredom by spending time alone in silence without constant stimulation.

These steps treat attention as something a person can shape through the surrounding environment. They also explain the title’s second instruction. Making your own algorithm begins with deciding what deserves repeated access to your mind.

The algorithmic majority

Odysseas sees the loss of attention as a symptom of a larger problem: the death of critical thinking. He turns to Alexis de Tocqueville’s Democracy in America, published in 1835. Tocqueville described the tyranny of the majority and the pressure that a newly enfranchised mass could place on individual judgement. Odysseas summarises the danger as a form of soft despotism created through agreement among the press, businesses, courts, police and the public. Elected representatives could then win support by appealing to popular desires rather than wisdom or virtue.

The recommendation feed creates a related pressure. It presents material that seems relevant and likely to earn a click, which makes relevance a poor substitute for usefulness or importance. The system shows a person more of what they have already chosen, so the first preference produces the next set of available preferences.

His example is a person who becomes interested in Marxism. The feed supplies Marxist memes, arguments, histories, debates and explainers until the person arrives in an imagined MarxTube, MarxGram or MarxTalk. Odysseas says the example applies to every community. A forum or website requires a person to decide to enter it. A recommendation system can push a person into a bubble before they understand that the bubble has become their whole field of view.

The narrowing reinforces confirmation bias. People tend to approach material that confirms an existing belief and retreat from material that challenges it. A worldview can become part of a person’s identity after years of use, so a challenge feels like a personal attack. The person responds by defending the belief or scrolling away. The resulting echo chamber repeats the same arguments and talking points, whilst anonymous posting adds opinions that may be careless, malicious or poorly reasoned.

Media ownership adds another layer. Odysseas names Qatar, Russia and Israel whilst arguing that ownership helps shape thought. He then returns to the remedy: self-doubt. This does not mean refusing to decide or sitting on the fence about every question. It means testing one’s own beliefs, seeking challenges and accepting that a conclusion can change. A person should reach conclusions for themselves instead of becoming a list of dogma and algorithmically supplied talking points.

Notes as a method of resistance

Odysseas gives one personal practice for keeping that distance. When he thinks a book, video or article contains something substantive, he waits until he can take notes before engaging with it. He watches low-quality debates for entertainment without asking much of them. When a formal debate, or an unexpectedly serious informal one, gives him something to work through, he takes out a notebook or stops watching.

Writing changes the relation to the material. Passive viewing lets an idea pass through the background whilst the person accepts its claims without testing them. Notes force the viewer to translate the idea into their own words and ask whether it is sound, logical and complete. Odysseas says that he sometimes writes down a position he initially accepts, then notices in the same sentence that something is missing or that he disagrees. The ink gives the thought enough distance to become an object of inspection. Changing one’s mind in that moment is his example of critical thinking in real time.

Becoming your own algorithm

The final problem is personal agency. Following or subscribing still leaves the feed in charge when the platform decides which items deserve attention. The common defence is that the algorithm works because it supplies more of what a person likes. Odysseas accepts the predictive success and questions the assumption behind it. People do not always click on what serves them. Outrage, drama, consumerism and fear of missing out can attract attention, just as a supermarket shopper can choose food that they know will leave them worse off.

When recommendation follows those impulses, culture flattens. Individual discovery becomes rarer, trends spread quickly, and a person can start to appear through an artificial version produced by the system. Odysseas’s answer is to become one’s own algorithm. He is purging his subscriber and follower lists, using Social Focus to block feeds and weaken recommendations, and going directly to creators, artists and writers whose work he respects. He compares this information diet to physical health: the mind takes shape through what it consumes.

The same principle leads him towards newsletters, printed books and newspapers. A newsletter lets a person choose a writer and receive that work directly. Going offline removes the feed’s intermediate ranking layer. Odysseas treats these older forms as practical ways to recover choice, then ends by asking viewers to share their own methods for dealing with algorithms.

Limits

The video combines a useful account of attention incentives with political claims that it does not document. The $400 billion valuation of Larry Ellison's wealth in 2025, the Ellisons' media acquisition, Oracle's 15 percent TikTok stake, the$ 26 million IDF donation, the $320 billion advertising forecast and the European regulatory response appear as source claims. The description contains no links to the reporting or data behind them. The caption also renders one reference to a political figure on Ellison’s private Hawaiian island as “big Yahoo”, so that detail remains unclear.

The five-stage description of recommendation systems is an accessible model rather than a technical account of a named platform. The video does not identify the source of its estimate that an approximate nearest-neighbour search might select 100 candidates, and it gives no experiment showing that addiction is the goal of every recommendation system. Its argument is strongest as a description of the incentives created by advertising-funded feeds. Its broader claims about who controls culture, how media ownership changes thought and how algorithms flatten culture remain interpretations that the video asserts rather than proves.