I’m Building an Algorithm That Doesn’t Rot Your Brain

Source: I’m Building an Algorithm That Doesn’t Rot Your Brain, New York Times Opinion, 7:10, uploaded 2025-11-19, playlist index 163.

Jack Conte opens with a blunt diagnosis: “Our brains are being melted by the algorithm.” He argues that the internet has moved from a place where following a person could lead to a sustained relationship with their work towards a system that fills the screen with whatever keeps a person watching. His proposed remedy is an algorithm built around people and creative work rather than advertising.

From following people to feeding attention

Conte describes attention as infrastructure. The phrase gives the argument its starting point. An algorithm sorts through the media produced online and decides what a person sees. For Conte, that formula exposes a company’s priorities more clearly than its public language does. He asks viewers to look at what a platform optimises if they want to understand what it values.

He attaches the current social-media experience to that choice. TikTok, Instagram and Facebook, he says, optimise for attention, which produces short-term fixation in place of long-term connection. Conflict and provocation become useful because they keep the feed moving. The result is a stream of doomscrolling, rage bait and polarisation. Conte also says that people feel lonely and depressed as this experience spreads. Those are claims made in the video. It provides no study, measure or comparison that would establish how much of those conditions the platforms cause.

The business model explains the direction. Advertising pays when a platform captures and keeps attention, so the algorithm turns attention into revenue for the company. Conte describes this as extracting attention from the brain and converting it into advertising income. A feed labelled “for you” therefore serves the platform’s commercial purpose before it serves the person looking at it.

Conte’s position carries an awkward qualification. He is the chief executive of a technology company himself. He co-founded Patreon, which lets creators share work with fans and earn money from it, and he has made music online for seventeen years with the band Pomplamoose. His criticism comes from inside the industry and from the perspective of a creator whose work has to find an audience.

He remembers an earlier internet through the experience of releasing music. When someone followed a person, the follower saw that person’s posts and could build a connection over time. The platform helped a community gather around the work. Conte says that the current feed has replaced that arrangement with a stream of whatever will keep the viewer’s eyes on the screen. The change affects the maker as well as the viewer because creative people now find it harder to build a community around a body of work.

The machine still has a job

Conte keeps recommendation systems in place. The internet contains too much material for a person to sort through alone, so some machine has to help select it. The question concerns the selection rule. A useful algorithm should help people find human creativity, while its success should depend on the quality of the relationship it supports.

He also insists that the change has to reach the working parts of the platform. A statement about community cannot sit above an unchanged advertising system and count as reform. The company has to alter what the algorithm measures, who receives the money and who gets to make editorial decisions.

Long-term relationships

The first principle is to keep attention in proportion and give long-term human connection more weight. Conte illustrates the idea through the creators a person already follows. A feed could surface that creator’s full catalogue instead of showing only the newest upload or the item most likely to trigger another click. Following someone would then expose the shape of their career and give the relationship time to develop.

The proposal treats a creator’s older work as part of the experience rather than as inventory that has already passed its commercial moment. It also gives the viewer a way to know a person through the changes in their work. Conte sets that against the constant sequence of rage bait, trend chasing and public dunks that makes the current feed hard to leave.

Money for creators

The second principle is to fund art rather than ads. Conte wants the algorithm to help creators earn money from their work, so the platform’s success would rest on producing income for the people who make the material. He accepts that an advertising company would see this as a difficult change. His answer is that the company should change its business to suit human well-being rather than asking people to damage their own well-being to suit the company.

Conte bases the proposal on the output he thinks a different incentive would allow. If creative people could receive reliable compensation, more art, podcasts, videos and books would exist because their makers could afford to keep making them. He says that this is what Patreon tunes its algorithms towards. The source gives the principle and the company’s stated direction, though it does not explain Patreon’s ranking system, payment formula or evidence that its approach produces the promised increase in work.

This is also where Conte’s position becomes a business proposal. Patreon benefits if creators and fans accept a platform that handles discovery and payment, so the criticism of advertising feeds doubles as an account of the kind of platform his own company wants to operate. That does not settle the argument. It does set a limit on how to read the video: the speaker is describing a public problem and presenting a model aligned with his company’s interests.

Human taste inside the feed

The third principle is to keep people in control of the recommendation process. Conte does not trust a machine with total editorial authority. He thinks an algorithm can develop better taste when people work alongside it.

His examples are small and practical. Creators can recommend other creators whose work they love. Superfans can receive payment for finding the best work from the artists they already follow. The arrangement gives a recommendation a visible human source and lets a person with deep knowledge of a scene guide discovery. Conte points out that curated discovery lists have existed for years. Twitter and Facebook once paid curators too, then abandoned the practice. He wants the payment and the role to return.

The proposal also changes what counts as expertise. A recommendation need not come from the platform’s central model or from a celebrity critic. It can come from a creator who knows the work around their own practice, or from a dedicated fan who has spent enough time with a small field to recognise what deserves attention. The algorithm can then distribute those judgements without pretending that taste emerged from a neutral calculation.

A platform that leaves room for art

Conte closes by explaining why he thinks the design choice matters beyond the feed. He says that societies benefit when they value artists because artists take risks, say things that people fear saying, bring people together, and offer unfamiliar ideas. Art can create memory and change a person’s view of the world. The internet could make room for those effects if its algorithms served people and their relationships with creative work.

The ending is a statement of possibility rather than a demonstration. Conte has described the incentives he wants to change and given examples of features Patreon might use. He has not shown a working system, a test result or a measure for “long-term connection”, “better taste” or a healthier experience. The video makes a case for choosing those goals. It does not establish that the proposed system can meet them at scale.

Limits

The video presents Conte’s account of how social platforms work and why their users feel worse. Its claims about loneliness, depression, polarisation and the effects of attention-based ranking remain source claims because the captions and description contain no supporting studies or links. The description mentions a 2019 New York Times Magazine account of the early hopes for the web, though it gives no article title or direct link, so it is not possible to identify that reference from the source material alone.

The three principles are clear enough to discuss, while their implementation remains open. A full catalogue feed still needs a way to order work. Creator and fan recommendations still need moderation and payment rules. A platform that funds creators still has to decide which forms of work qualify and how much each contribution earns. The video leaves those decisions outside the frame.