P O T E S T # 1

* * *

A small summary in our first collaborative observation of the Pornhub algorithm.
( final report )

January 19th 2020 we openly asked to



Add the poTREX extension

What is pornhub.tracking.exposed?


Follow the links

NSFW ;)


Unravel the algorithmic mist

our manifesto

Add the poTREX extension

What is pornhub.tracking.exposed

we analyze platforms

* * *

NOT PEOPLE

TREX browser extension collects data on a voluntary basis, in order to decipher the functioning of the proprietary algorithms for public interest.

ANONYMIZATION PROCESS

  • 01. Unique and secret token

    Every browser has a unique code attributed

  • 02. Your choice

    Total control managing the data provided: visualize, download or delete

  • 03. No context

    No data about where you are coming from, who your friends are or what they are doing and such

  • 04. Not our customer

    We are not obsessed by you ;)

Follow the links

NSFW ;)

research protocol

* * *

EVIDENCE

Navigate the website following our instructions to join the experiment.

  1. Open the PornHub homepage
  2. Open the page generated by Pornhub containing the videos recommended for you.
  3. Open this video, it's been on Pornhub for 11 years.
  4. Open to recommended for you, let's see if and how they changed.
  5. Open this video, which has been published yesterday .
  6. Lastly, open the homepage.

Observations

  • Homepage: it keeps changing even for unlogged users, let's record evidence
  • Recommended: where the personalization algorithm takes action
  • Videos: we compared one video uploaded 11 years ago with a 24 hours old one

Unravel the algorithmic mist

our manifesto

a closer look at the data

* * *

.CSV RELEASE

Follow up through some announcements on the website:
In one day, we posted our first updates
After three days, the estimated amount of evidences collected
On the 25th, our first result and the second version of the dataset

THE DATASET

* * *

Each entry represent a suggested video from Pornhub: each video snippet you might click on.

"Type" and "Steps" identify the phase of the test:
homepage returns 52 videos
recommendation page 25
video pages 8

Only sessions by a contributor who did the 6 steps in sequence are considered: 87

{
        "title": "Sunny Sextape on the Sofa! Squirt, deepthroat, prone bone - Amateur LeoLulu",
        "authorName": "Leolulu",
        "authorLink": "/pornstar/leolulu",
        "duration": "17:15",
        "href": "/view_video.php?viewkey=ph5e18b11299830",
        "savingTime": "2020-01-19T22:18:10.522Z",
        "metadataId": "738c411c67c7b6107bbb3ff8631070011a814f48",
        "clientTime": "2020-01-19T22:17:48.000Z",
        "size": 421227,
        "randomUUID": "INITucmr5condtj2zkfy9o6cv4",
        "type": "home",
        "processed": true,
        "step": 0,
        "session": 1,
        "pseudo": "blueberry-cake-pistachio",
        "sectionName": "Hot Porn Videos In United States",
        "sectionHref": "/video?o=ht&cc=us",
        "sectionOrder": 0,
        "displayOrder": 0,
    },
        

F I N D I N G S

* * *

A small summary and next steps

potest #1

* * *

WATCHING THE FIRST VIDEO

(Uploaded 11 years ago)

All participants got exactly the same 8 related videos.

potest #1

* * *

WATCHING
THE SECOND VIDEO

(Uploaded 24 hours earlier)

Participants got different related videos, depending on the time of the day they performed our poTEST.

potest #1

* * *

WATCHING
THE SECOND VIDEO

(Uploaded 24 hours earlier)

Animated timeline: evidences recorded between 00:00 and 24:00

potest #1

* * *

RECOMMENDED PAGE

A few videos are different after the procedure

Watching just one video could not be enough for a strong personalization in the /recommended page


videos that appear on both evidence
new videos

potest #1

* * *

HOMEPAGE

New videos suggested after the procedure

We still don't know the reason for these changes, but the first and the second homepage visualizations got different videos after a few clicks.


videos that appear on both evidences
new videos

What we didn’t find out,
but we’ll keep pursuing

* * *

  • Does any particular producer benefit from an advantageous treatment from the algorithm?


  • Does the recommended page change for non-logged-in users accordingly to what has been seen? How?

K N O W M O R E

* * *


Final Report

The CSV


Did you like it? Contact us!

Thanks to an input on reddit we look at the first insight and defined how to confirm the thesis, how to test new conditions, and in the meanwhile look if our guesses are true also in the Gay section of PornHub! We want to keep this analysis and the formulation of the research question open, which is true also for future tests. If you want to reach out, you are welcome to join our community.