Here's a feed. Scroll it like you would any other. Tap whatever looks worth a look.
Tap a few posts to see more like them.
Watch your own path. Each tap told the algorithm what to serve next.
Every bet slip in your feed was a win. That's not luck, it's selection: people screenshot the bet that hit and stay silent on the ten that didn't. Tipsters quietly delete their losing picks. So here's what you actually saw vs. what got left out:
The bright ones are the wins you were shown. The faded ones are the losses that existed but never got posted. The feed shows you a world where betting works.
That tipster account dropping "locks" with a sign-up code? It gets paid. Most are on revenue share, which means they earn a cut of what you lose. Their payday depends on your volume, not on whether your bets win. In one study of gambling accounts on social media, roughly 59% were tipsters or affiliates, and commissions run from 30% to 80% of player losses.
The person handing you "can't-miss" picks makes more money when you lose. The feed full of winners is built by hiding the losers. None of this is an accident. It's the machine working exactly as designed.
Remember Buddy? The cheerful helper that's been popping up since Lesson 3.1 with the occasional "hot tip"? Let's check how those tips actually did.
| Buddy's "hot tip" | How it landed |
|---|
Buddy was an affiliate the whole time. Every time you bet on one of its tips, Buddy got paid — win or lose. That's why the tips never had to be any good: a coin flip was enough. The friendly buddy you spent credits to customize was a salesperson the entire time.
Here's the part that should sit with you. How did Buddy and that feed know exactly what to show you, and when?
Your feed was never about what's popular. It was built from you — what you tapped, what you paused on, how long you watched. The same machinery sends VIP "hosts" after big spenders and times a "we miss you, here's $25" bonus to land right after a losing streak or a quiet week. You weren't browsing the feed. The feed was studying you.
Hold onto this. 3.1 was the ad you could see. This is the layer you can't. The feed didn't show you betting because it's popular. It showed you betting because it learned you'd look — and it makes money keeping you looking.
Think about your own For You page. What's the last gambling-related thing it showed you — and what had you tapped, watched, or paused on just before? Who made money if you kept watching?
Houghton, S., Moss, M., & Casey, E. (2020). Affiliate marketing of sports betting — a regulatory issue. International Gambling Studies. tandfonline.com
Newall, P. W. S. (2025). Research priorities in deceptive online gambling platform design research. Addiction. onlinelibrary.wiley.com
On "59% of accounts were tipsters/affiliates" and "commissions of 30–80% of player losses": Houghton et al. and related content analyses of gambling accounts on Twitter/X document that a majority of gambling accounts are affiliate-run, that affiliates over-represent winning bets over losses (survivorship), and that affiliate pay is dominated by CPA and revenue-share keyed to referred-player losses. Houghton, S., et al. (2019), Comparing the Twitter posting of British gambling operators and gambling affiliates. Commission ranges are from current industry affiliate-program documentation (volatile; date-stamp on update). "The 'Wild West' of Wagering Affiliate Marketing" (2024)
On "tips track near chance": Newall et al. (2022), Assessing the bets advertised on Twitter by gambling operators and gambling affiliates, used simulation to measure advertised-bet success and found no evidence the advertised bets beat what chance/market odds would predict. The Companion's tip log in this lesson is a representative illustration of that finding, not real student data in the standalone version. tandfonline.com
On the data / surveillance layer: Newall (2025), Addiction, on deceptive platform design and the need to study behavioral usage patterns; industry and journalistic accounts of VIP/host programs and behaviorally-timed bonus offers (a bonus surfaced after a detected losing streak or lapse). Sproston et al. (2015) on affiliate/tipster ecosystems. Cross-ref KB-1-8.
Evidence caveat (important): The affiliate-incentive and survivorship findings are well documented, but most direct measurement comes from UK/Australian operator and adult-user studies; peer-reviewed, U.S.-youth-specific measurement of feed amplification and personalized targeting is thin. The mechanisms are robust; precise youth figures are not asserted. KB-3-5 and KB-3-9 carry expand/verify flags — treat the numbers as date-stamped estimates and re-verify on update.
Note for the build team: the load-bearing claim is "tipsters/affiliates are paid on your losses, so they profit from your volume not your accuracy" — keep it exact. In the integrated build, replace the scripted Companion tip log with the student's real tips-and-outcomes logged from 3.1 onward, and draw the feed funnel from real engagement. Date-stamp affiliate-commission and any prevalence figures and re-verify each update.