#recommendation algorithms

[theemperorsfeather]

I don't mind letting your programs see my private data as long as I get something useful in exchange. But that's not what happens.

This is a really interesting essay about advertising trackers and recommendation algorithms (and why they are terrible) and so on.

A former co-worker told me once: "Everyone loves collecting data, but nobody loves analyzing it later."  This claim is almost shocking, but people who have been involved in data collection and analysis have all seen it.  It starts with a brilliant idea: we'll collect information about every click someone makes on every page in our app!  And we'll track how long they hesitate over a particular choice!  And how often they use the back button!  How many seconds they watch our intro video before they abort! How many times they reshare our social media post!

And then they do track all that.  Tracking it all is easy.  Add some log events, dump them into a database, off we go.

But then what?  Well, after that, we have to analyze it.  And as someone who has analyzed a lot of data about various things, let me tell you: being a data analyst is difficult and mostly unrewarding (except financially).

See, the problem is there's almost no way to know if you're right.  (It's also not clear what the definition of "right" is, which I'll get to in a bit.) There are almost never any easy conclusions, just hard ones, and the hard ones are error prone.

[shacklesburst]

We make a lot of fun about Youtube’s recommendation algorithm, but there’s at least one thing it’s been getting better at: recognizing when you want to play music and which type of music you want to play. Not even a year back, having Youtube open in the background starting with some song and then autoplaying from there might’ve led to some “normal” video by somebody you’re subscribed to, or a funny video compilation if some were in your recent history somewhere along the line.

This basically doesn’t happen anymore. It nowadays always stays with music, mostly with known artists you like, only branching out tentatively within the same genre, basically. The trending or home pages basically don’t ever yield anything interesting for me (other than showing recent uploads by subscribees), but the next song selection algorithm really does its job now.

[canofcherrycoke]

trying to outrun the mortifying ordeal of being known by outsourcing your one human task of self-creation to AI is not less embarrassing than just having cringey phases btw. in fact it may just be more embarrassing

[entertainment01281433226]

{

"@context": "https://schema.org",

"@type": "Article",

"headline": "",

"description": "",

"articleSection": "entertainment",

"keywords": "\"Behind the Scenes: How Streaming Algorithms Shape Your Favorite Shows and Movies\"",

"datePublished": "2026-02-02T22:17:01.899943",

"dateModified": "2026-02-02T22:17:01.899943"

}

[katmh]

I found it both freeing and disturbing to spend time on a platform that didn’t ask me to pretend that I was on the Internet for a good reason. I was not giving TikTok my attention because I wanted to keep up with the news, or because I was trying to soothe and irritate myself by looking at photos of my friends on vacation. I was giving TikTok my attention because it was serving me what would retain my attention, and it could do that because it had been designed to perform algorithmic pyrotechnics that were capable of making a half hour pass before I remembered to look away.

Jia Tolentino, “How TikTok Holds Our Attention”, The New Yorker https://www.newyorker.com/magazine/2019/09/30/how-tiktok-holds-our-attention

[katmh]

TheirTube

How do the recommended videos look like on their Youtube?

[open-source-sorcerers]

Sistemas de Recomendación en el Open-Source

Los sistemas de recomendación es un sistema implementado en un software que recomienda o predice el contenido que el usuario es propenso a ver para mantenerlo enganchado en el uso continuo de la plataforma. El uso más común de este tipo de algoritmos son en las plataformas de streaming como Netflix.

Logotipo de Netflix

Cuando terminamos de ver algún contenido en alguna de estas plataformas, una vez que inicien los créditos, ella nos arrojará algunas selecciones de títulos que, basado en nuestro historial de reproducciones, podrían ser de nuestro agrado. De esta forma, nuestro uso y consumo de contenido es imparable, esto es a lo que Nick Seaver, autor del texto “Captivating algorithms: Recommender systems as traps. Journal of Material Culture” llama una trampa.

Mientras que el autor lo llama de esa manera, en inglés ha adquirido un término propio, “Binge”. Incluso, en el idioma español utilizado en México, hemos adquirido el anglicismo como “bingear”. Aunque comúnmente, este término está relacionado con la borrachera, su uso es más popular para los maratones de series o películas que proponen las plataformas de streaming gracias a su sistema de recomendación.

Imagen del video “Why We LOVE Binge Watching And What It Does To Your BRAIN” por The Friendly Brain en YouTube.

Recientemente ha habido una alza en el uso de sistemas de recomendación gracias a estos nuevos medios. Debido a su popularidad, muchas alternativas “Open-Source” han salido de forma gratuita para su implemento en distintas aplicaciones, tal como lo es “Raccoon Recommendation Engine”.

¿Qué es lo que esto significa para los softwares, sitios web y, sobre todo, para nosotros?

Cuando un sistema se vuelve “Open-Source”, los programadores y diseñadores son capaces de estudiarlo. Es similar a cuando una persona abre a un animal para estudiar la forma en la que su cuerpo funciona. De la misma manera que esto genera avances en la medicina veterinaria respecto a cierto tipo de animales, el código abierto también abre las puertas a un avance tecnológico dentro del diseño digital y la programación. Los algoritmos son libres, para que todos los usen, implementen, aprenden de ellos y los conviertan en algo mejor.

Esto significa que, los sistemas de recomendación, empiezan a ser utilizados en plataformas y softwares que no se habían visto antes. Buscan enganchar a su usuario cada vez más en este medio digital, logrando que consuma continuamente en su propia plataforma, sin dejarlo escapar a otra. Lo atrapan de la misma forma que Seaver cita a Mason “Una trampa es una invención cuyo propósito es inducir a los animales a cometer encarcelación, auto arresto o suicidio”, atacando nuestra propia curiosidad por consumir contenido.

Seaver, N. (2019). Captivating algorithms: Recommender systems as traps. Journal of Material Culture, 24(4), 421–436.

Sarmah, H. (2019). 5 Open-Source Recommender Systems You Should Try For Your Next Project. Retrieved 1 May 2020, from https://analyticsindiamag.com/5-open-source-recommender-systems-you-should-try-for-your-next-project/

[bluepiposts-blog]

Recommender system has the effect of guiding the user in a personalized way to interesting objects.Collaborative,Content and Hybrid are most common systems.

In this blog, different types of recommender systems are classified which serves as building blocks for different systems.There are majorly six types of recommender systems which work primarily in the Media and Entertainment industry, read on to find out about them.