Ricochet Frame

UPD: I somehow spent all day arguing about this shit, so I let my autism roam free and actually went through all 1606 reviews to see how big this "speedrun refund" crisis really is. Out of 1606 reviews, only 35 are from people who refunded, and exactly five fall into the speedrun-refund category. Five. Not dozens, not thousands, as many people claim here. Only 5. Here they are: 1. -s***: "GREAT GAME, finished within 1:40 hrs (refunded)" 2. D1**: "1시간 30분 환불런 성공" (1 hour 30 minutes refund-run success) 3. 정약*******: "2시간 올클리어 기준으로 환불런이 가능..." (with a sub-2-hour all-clear, a refund run is possible...) 4. _p**: "환불런 해버렸음" (ended up doing a refund run) 5. de****: "beat ts game fast enough to refund it, me and gang locked in fr ✌️" Now let's do the actual math and estimate how many of these 55k refunds are abusers. For that, let's look at the baseline: Total refunds (R) = 55k Total copies sold (N) = 262k (based on 21% refund rate) First, a quick reality check: the achievement for beating the game in under 2 hours was only earned by 7.1% of players. So: 262k sold x 7.1% = 18602 players max. This means 18602 is the absolute hard ceiling for speedrun refunds. You physically cannot be a "speedrun-refunder" without actually beating the game in under 2 hours first. But that's just the ceiling, not the real estimate. Now let's look at the reviews. We have 35 refunded reviews, and 5 are clear speedrun abuses. That's 14.3% of the refunded reviews. But you can't just blindly apply that 14.3% to the entire 55k refund pool. Why? Because writing a review isn't random. A guy who wants to flex "I finished this under 2 hours and got my money back lol" is obviously exponentially more likely to leave a review than a normal person who simply disliked the game, felt it was too hard, or had technical issues, and quietly refunded it without saying a word. We need to account for this review-writing bias. Let's call it lambda (λ), meaning how many times more likely a speedrun-refunder is to brag in a review compared to a normal refunder. The refunded-review sample has 5 speedrun-refund reviews and 30 normal refund reviews. So the observed ratio is: [speedrun reviews] : [normal refund reviews] = = 5 : 30 = = 1 : 6 Now, let: A = actual number of speedrun refunds 55k - A = actual number of normal refunds p = probability that a normal refunder leaves a review λp = probability that a speedrun-refunder leaves a review So the expected review ratio is: (λpA) / (p(55k - A)) = 1 / 6 The p cancels out: λA / (55k - A) = 1 / 6 Then solve it: 6λA = 55k - A 6λA + A = 55k A(6λ + 1) = 55k So the formula is: A = 55k / (6λ + 1) Now let's plug in different assumptions for how loud these flexers are: - If λ = 1 (naive assumption, zero bias): A ~ 7,857 cases. - If λ = 2 (twice as likely to review): A ~ 4,231 cases. - If λ = 5 (five times as likely to review): A ~ 1,774 cases. - If λ = 10 (ten times as likely to review): A ~ 902 cases. The naive projection assumes both groups review at the exact same rate, which is a completely unrealistic assumption when dealing with internet flexers. Let's do another sanity check with the overall review rates. The overall review rate for the game is 0.613%. But the review rate for refunded users is a microscopic 0.064%. Regular people who refund just don't leave reviews. If those 5 speedrun reviewers are sampled at the normal review rate of a regular, non-refunded buyer, which is 0.759%, it scales to just ~659 total cases. If they review at the general average rate, it's around ~815 cases. So, what's the actual takeaway here? The projection is roughly: - ~650-900 cases if speedrun-refunders review like normal players. - ~1,000-2,000 cases if they are a few times more likely to post about it. - ~7,900 only under the completely naive and unreasonable assumption that there is zero review bias. From the available data, the problem is massively overstated. It's clearly not 55k abuse cases, as Mateo tries to claim here. The somewhat reasonable estimate that it's likely somewhere between a few hundred and a couple thousand clear speedrun-refund cases max for this game. Of course, this is all without accounting for how common speedrun-refunders are on Steam in general, which would push the estimated number even lower. But I don't have the energy to collect that data, and I think the numbers already speak for themselves even at this point. Finita la commedia, thank you.