Noise about signals: Information Cascades, GIGO, and You

Maybe Nerds Were Supposed to be Bullied

Patrick Chapin is definitely one of those nerds, with one of the most impressive writing careers in the TCG space. Years ago, Chapin saw a vision of the future where TCG metas grew inbred and static in a vicious cycle of self-fulfilling prophecy (editor’s note: google “hyperbole”). Information Cascades, as he called them, emerged from the tendency of many to take performance data as a direct indicator of strength. Chapin contended that many players simply copy the lists which performed well, resulting in a “cascade” wherein each subsequent individual was blindly trusting the list before theirs. Like lemmings, gamers would simply pick the list that last performed one after another, resulting in an ecosystem devoid of new material and innovation.

We find ourselves now surrounded by data analysis. Gamers pore over spreadsheets for The Truth, safe in the knowledge that numbers don’t lie. Anyone who has experienced a “um actually… my deck is 54.36% into yours so you’re better off saving us both time by conceding,” should already know that data analytics in TCGs can often be a chimpanzee-with-machinegun style arrangement. I argue that, yes data analysis is a uniquely powerful tool in the arsenal of any competitively minded gamer, but we owe it to ourselves to be thoughtful about how we engage with data. There is no harm inherent in trusting the work done by others. However, stale metagames and predictable behaviors are exploitable and potentially dangerous for format health. I believe that we are on the precipice of a data analysis influx that threatens to poison deck discussion and card choice strategy.

In With the Garbage, Out with the Garbage

Anyone that handles data will know about GIGO, or garbage in, garbage out. GIGO is why Frank Karsten gets mad at the jokers who register their deck as “4c Zuberas” or some other nonsense (a problem FaB avoids entirely by not even having deck names, well done LSS). Simply, if the data going in is no good, analysis of that data will yield nothing of value. This is the principle behind popular Magic aphorism “leagues are noise,” low-entry, low-stake leagues result in metas that deviate from their more serious counterparts. If the results coming out of MTGO leagues were treated with the same weight as results in the Pro-Tour, we’d be doomed.

Data analysis is only as good as its dataset, and we can imagine datasets in TCG analysis as individual ecosystems. The larger and more diverse an dataset, the more useful its metrics. Since TCG data analysis orbits the organized play structure, performance metrics must be analyzed within their metagames. Accordingly, there are many factors outside of gameplay that can influence the construction of a given metagame and the performance of the decks within. Consider the recent Magic World Championship, notable for disparate performances across the same lists. Seth Manfield’s green black deck seemed only to behave in his hands, with many of his peers floundering on the same 75. Similarly, Kano extraordinaire and fellow Cold Foil Cartel member, Manu, made day 2 with Kano in Portland, a feat he alone accomplished.

Not only is data only as useful as its quality, but performance metrics also often neglect the pilot’s position. If he had made his deck decisions in line with the data alone, Manu would have been playing a deck he didn’t care for with cards he didn’t choose. His example speaks to the value in thoughtfulness, that performance in TCGs does not just mean imitating those who do perform well. He and those like him will find themselves performing better than average even in hostile metagames.

What FaB Does Right

Part of what has been fascinating about my journey into Flesh and Blood is the tension between wanting to specialize and needing to specialize. Card prices can be prohibitive, and access will always be limited, tempering the temptation to simply bring last week’s winning deck to the Armory. While I am not certain that FaB’s heavy investment model is good for the health of the game, it does sometimes force its players to improve.

Innovation like Vaporize // Shock in Aurora will never result from staring at spreadsheets, it will only ever come from time spent playing that hero. In my estimation, Flesh and Blood is still in its infancy. With so much room left to grow, I urge that those who do care to be among the vanguard lead by example. Aim not to play deck that is winning, aim to be the one who breaks the meta.

You

Information is cheap in the internet age. Anyone can make a spreadsheet, and anything can be data. It is up to us to hold ourselves to higher standards. If the average TCG grinder spends too little time considering the card choices in their list, their ill-conceived decks will smash into one another, producing vulgar data for vulgar consumers. Don’t allow yourself to be among their ranks.