Meet The MIT Grad Who Cracked The Code For MTV’s ‘Are You The One?’
Part of what makes MTV’s Are You The One? such an amazing reality TV show is the drunken bickering, sure, but the far more compelling part is the surprisingly large amount of math that’s going on behind the scenes as the house frantically scrambles each season to discover their perfect matches.
Considering that each season begins with more than 3 million potential combinations that could happen when you tell 10 men and 10 women in the AYTO house to find their soulmate in a matter of a few Red Bull-and-tequila-fueled weeks, that’s a pretty gnarly statistical conundrum to winnow down. We’ve been working for the past four seasons on our own matchup grid to figure out the house’s perfect matches, but by far the best expert around is Nick Uhlenhuth, who has been keeping the Wikipedia page for the show updated every week with the odds of each member in the house being somebody else’s perfect match.
It’s an incredible public service, and one we’re all eternally grateful for. Even better, we had the chance to catch up with Nick recently, and learn a little bit more about him and the work he’s been doing. Check it out, and be sure to check out his work each week too!
Can you give us a little info on your background?
I grew up in Kentucky and have always been interested in technology. I just graduated with a Computer Science degree at MIT, and I’ll start my job at Microsoft in a few weeks!
How did you get interested in Are You The One?
I was on vacation with some friends in Florida, and we decided to watch some TV. All the sudden, some dude named Ryan Devlin starts talking about beams of light, and we’re hooked. The rules of Are You The One instantly reminded me of a childhood board game called ‘Mastermind,’ so there was some nostalgia.
Can you walk us through your model — how you built it, and how you produce estimates of each couple’s match probability?
In actuality, there is no estimation involved – it’s all just numbers. The percentage for each couple is simply the percent of possible remaining matchings that contain that couple.
At the start of the season there are 10 guys and 10 girls. This means there are 3,628,800 possible ways to match the guys with the girl. The first guy can choose from 10 girls, the next guy can choose from the remaining 9, the next guy can choose from the remaining 8 girls, etc. (10*9*8*7*6*5*4*3*2*1 = 3,628,800). Ok, sweet.
Each week, we get information from Truth Booths and Matching Ceremonies. We use this information to remove ‘illegal’ matchings from the 3,628,800 possible matchings. As we continue to remove illegal matchings, the number of possible matchings gets smaller and smaller. Clearly, if we can reduce the number of possible matchings to just 1, then we can win the game with 100% certainty.
How do we reduce the number of possible matchings?
If a couple is confirmed Not a Match, then we know that any matching containing that couple is an “illegal” matchings. So, we remove all matchings that contained that couple. For example, in Week 1 we know Prosper and Tori were Not a Match. So, from the 3,628,800 possible matchings, we remove any matchings that had Prosper matched with Tori!
Similarly, if a couple is confirmed a match, then we can remove all matchings that don’t have that couple matched. For example, in Week 3 Cameron and Mikala were a Perfect Match. So, I removed any matchings that didn’t have Cameron matched with Mikala!
These are a bit trickier, but follow the same basic idea as the Truth Booths.
After Week 6, we have already reduced the number of possible matchings from 3,628,800 to 8! In other words, only 8 matchings are consistent with the Truth Booth information and the Matching Ceremony information!
Can you explain how you derive the percentages?
The calculation is much less complicated than you might think. The table of percentages that I put on Wikipedia simply indicates what percent of the remaining possible matchings have Contestant A matched with Contestant B. After Week 6, there are only 8 possible matchings. If 7 of the 8 matchings have Contestant A matched with Contestant B, then the table will show 7/8 = 88% (I round to the nearest whole number). Notice this is the case with Camille and Asaf right now!
In other words, the percentage table just looks at an individual couple’s likelihood of being a perfect match.
Things get a little more involved when we consider which complete matching is the most likely. To do this, I assign a score to each of the possible remaining matchings (there are 8). This score is calculated by summing up the percentages assigned to each couple in the matching. A higher score indicates a higher likelihood of being correct. These predictions become much more accurate each week. Last season, I predicted the correct matching after the final Truth Booth! Here are the scores assigned to the remaining 8 possibilities: I don’t currently put this on Wikipedia, but I could if people want!
What made you want to publish your odds calculations on the Wikipedia page for the show, rather than as a private blog? Have you gotten any feedback or good interactions with other fans/contestants from the show?
Wikipedia is one of my go-to sources for information (don’t tell my high school teachers!). While watching the show, I went to Wikipedia to see if there was any information. Sure enough, there was a “Progress” section for Are You The One! But I realized I could add some useful information that would be almost impossible to create without a computer. That night, I coded for 4 straight hours and created the percentage chart that can now be found on Wikipedia. A couple days later I decided to add some color to the chart to make it more readable. (My code can be found here.
I chose Wikipedia because I figured that many other AYTO fans would be able to access my data. The upside is my Table can reach a lot of people. The downside is that sometimes trolls change the numbers in the table or update the table incorrectly!
Because my name isn’t really attached to the table on Wikipedia, I haven’t had anyone reach out or give me feedback. However, I have had several people ‘clone’ my code from my github repository (linked above)! That means people essentially have downloaded my code to their computers so that they can generate the table themselves. (I guess some other nerds watch this show!)
How do you think you’d fare in the AYTO house?
I think I would end up being one of those dudes who tries to work with logic rather than going with Devlin’s advice of “follow your heart”! I don’t think I would get caught up in much of the drama, and I’d try to lead strategic efforts.
What do you think is the optimal strategy for a contestant in the AYTO house?
Initially, I would definitely encourage some sort of speed-dating thing so that each contestant can get an idea of who is the most compatible.
I think the best strategy is to put couples that are closest to 50% into the Truth Booth. Remember, this means 50% of the possible matchings contain that couple, and 50% do not. This way, no matter what the Truth Booth says, you are inevitably able to cut the number of remaining possible matchings in half!
Another key point of my strategy would be to not put extremely likely matches into the Truth Booth. For example, I would have never put Sam and Alyssa in the booth. Instead, I would just always have them sit together at the ceremonies to ‘guarantee’ there isn’t a blackout.
ed. note: This is the exact opposite of what we suggested Sam and Alyssa do after week 5, because we’re very dumb and Nick is very smart.
Who do you think would perform better, you as the host of AYTO or Ryan Devlin as an MIT undergrad?
I think I would be a pretty good AYTO host. I’m not sure what sort of behind-the-scenes work it entails, but I think it would be fun to come up with challenges and some twists. I think the biggest issue is that I’m only 22, so I doubt the contestants would take me too seriously!
I have no idea how Devlin would fare at MIT, but it would probably make for an entertaining show.
What’s been your favorite part of the show so far?
In this season, my favorite part was probably when Tori and Morgan drew all over John while he was sleeping.
Who do you think has been the most strategic contestant in the show’s history?
Unfortunately, they don’t show much of the strategizing, so it’s hard for me to know who is responsible for the progress. That being said, Devin was pretty convincing with his Red Solo Cup model during Season 3. 😛
Have you thought about using the AYTO model to determine your own perfect match?
I’ve already found mine 😉