1,000 Greatest Drivers: Rick Kelly

The best things in life aren't quantifiable.

I reminisced a bit about tournament Scrabble in my Red Byron entry a month ago after I mentioned a long comment I made in the comment section of Stefan Fatsis’s first Substack post. I was reacquainted with Matthew O’Connor for one of the first times since I quit Scrabble tournaments in 2017 and he had just launched his own Substack for Scrabble strategy a few days before and we are now both following each other. When I was playing tournaments, he was the best Scrabble player in Syracuse and I was second-best even though he was 14-19 years old at the time and I was 27-32. He no longer lives here so I guess I’d once again be the best Scrabble player in Syracuse if I cared about it anymore which, of course, I don’t. We had other stuff in common too. He’d had third and fifth-place finishes in the Syracuse Regional Spelling Bee in 2010 and 2012; I’d finished sixth in 1999 and actually uploaded it. We first met at a club tournament in 2012 at the now-abandoned ShoppingTown Mall, which was right around when he first became an expert (he later won the World Youth Scrabble Championship shortly after I quit). I had no issues losing to a kid as I’d been that guy myself in Scrabble-by-Mail when I was a kid, but I must admit he deeply irritated me when after one of my early tournaments, he told me I was “more suitable for a casual level of play” at a time he was exceedingly brash and confident while I was morose and down on myself about everything in life, a feeling that has never abated since. I can’t remember even liking myself since 2002 at the latest. It certainly motivated me and I think I proved him wrong. I could beat him and I arguably even overachieved against him, going 7-14 and gaining rating points against him. I played him more than anyone else in tournaments.

Anyway, he just released a strategy book and I wanted to review it and riff on it even though I doubt there’s much crossover between what I’m doing and what he’s doing. His was not the first Scrabble strategy book I’d ever read. I’d read Joe Edley’s Everything Scrabble, Joel Wapnick’s Champion’s Strategy for Winning at Scrabble (I think it was that one - he wrote three), and Kenji Matsumoto’s Breaking the Game and Words of Wisdom. I admit I found it a little goofy that he named it How to Play Scrabble Like Matthew O’Connor because none of those other guys named their books after themselves, but he told me he didn’t mean it to be self-aggrandizing and he certainly acknowledges several times in the book that Nigel Richards is the goat and he does not claim to be.

Matthew’s book had quite a bit in common with Kenji’s. Kenji was the first person I read who explained why having a particular style isn’t good by arguing that an offensive player is right some of the time, a defensive player is right some of the time, but an optimal player should make an offensive play when the board situation calls for offense and a defensive play when the board situation calls for defense. This was a much bigger emphasis in Matthew’s book than Kenji’s. Once I read Kenji, that idea instantly clicked with me and I got it and agreed with it, but it also made the game honestly less interesting to me personally, not more, and I’ll get to that later.

Matthew focuses more on defense than other players so a lot of people see that as his “style”. He does not see it that way and just sees himself as choosing optimal plays and thinks that other players don’t pay as much attention to defense as they should. Honestly, I agree with that too. A lot of Scrabble experts determined the best play simply by calculating/estimating the “equity” of the play, which is the sum of your score and the expected value of the tiles you keep rather than an average leave. From what I remember (and maybe it’s changed) you’re expected to score about 25 points more with a blank than if you don’t have it, 8 more with an S, 4 more with an E, X, or Z, while Q used to be -13 before the words QAT and QI were added and now I think it’s -5 or something. And then sometimes if you put certain sets of tiles together they’re worth more or less than the sum of the tiles on your leave combined depending on whether they’re “synergistic” or not (frequently appear in words together). I never really bothered to get in the weeds with that personally and played on vibes more than that. I didn’t feel it was worth memorizing that shit because I knew these were just averages and obviously certain tiles would be worth more or less depending on whether the board is open or closed to different types of plays. I also thought personally that would be a waste of my time because I had bigger issues than that. I think my strategic instincts were always pretty sound, but my biggest problems were just that I was a lazy word studier (like I said, I had the lowest average bingo probability of any player ever, which obviously means my strategy was better than my word knowledge) and I was also not very good at seeing my options. If you gave me a multiple choice test between several candidate plays I could make in a given board scenario, I think I’d do pretty well. But I often struggled to see strong defensive plays, like 22-point 5-letter words that make 3 overlaps or something like that. I know those kinds of plays are the best plays a lot, but I just often didn’t see them and sometimes made dumb-looking plays because I couldn’t see my good options. Maybe if I’d been able to see my options, there would have been a point for doing equity calculations.

But I digress. The real issue is that equity does not factor defense and it simply measures what will maximize your point total when the winning percentage of your play is what you should be looking at. Apparently, a lot of players focused on maximizing points instead of maximizing winning chances. In some environments, that could be good. I know back in the ‘80s, British tournaments were decided by who scored the most points instead of wins or losses, so players actually worked together to open the board and create great opportunities for themselves so both players could maximize their scores as much as possible. You know what? I think I’d find that more fun. But American tournaments were never like that and the Brits gave up on collaborative games once the World Scrabble Championship started in 1989. You could certainly adjust equity calculations to factor in how many more points your opponent is likely to score based on different board positions, but that’s a lot murkier and harder to calculate than points + leave, so I guess a lot of experts defaulted to that, and I think sometimes my more vibes-based playing was actually even superior to that.

Matthew, like Kenji, put a lot more emphasis into board geometry and shaping the board so it will suit you than Joe and Joel’s books did. Matthew expanded that idea by arguing that players frequently should consider preemptive defense by blocking important lines earlier in the game before they even become relevant, and that wasn’t something I’d thought about much before. He also argued that people shouldn’t play Scrabble like chess because chess has complete information and Scrabble has incomplete information. That was never a problem for me since I was always into Scrabble first and I never played chess seriously, but a lot of people play both and I agree you’ll mess up if you try to adapt chess-based thinking to Scrabble except in the endgame when all the tiles have been drawn.

I agree with all that. I get it. Does it really make me have a hankering for wanting to play Scrabble again? Honestly, no. I agree that what Matthew and before him Kenji advised maximizes winning chances. I just don’t find it intellectually interesting. It’s really weird for me to be saying this as someone who’s trying to do auto racing sabermetrics, but I really kind of agree with the people who argue analytics has hurt sports. If variety is the spice of life, optimization to maximize the chances of winning genuinely makes things more boring aesthetically. Most players in a post-analytics world are conditioned to play in the same way, so there’s ultimately less variation, less creativity, and in my opinion, less fun. Every baseball player is swinging for the fences to hit homers so we’ll probably never see a Tony Gwynn again. Every basketball player is now focusing on 3-pointers so there’s less variance in strategy and most games look fundamentally the same. It feels like games are more fun before they’re optimized. The same held true in auto racing. People look fondly on the ‘60s as the greatest era of the Indy 500 because there were so many cars that looked wildly different until teams converged on the most aerodynamic types of chassis after which point almost everything looked the same, and it’s telling that that is when the gearheads started to lose interest. Sure, in racing, you can still have different-looking cars especially in sports car racing (which has always valued that), which is obviously why they’ve instituted all those balance of performance requirements to ensure different-looking cars have similar speeds.

Once you’ve figured out in any game or sport what the optimal strategy is, there’s no turning back. Players ultimately train themselves to be something akin to human robots or AIs making the calculated correct moves with no room for creativity because it’s suboptimal, and I guess that world interests me less and less. I realize attempts to play Scrabble with chatbots have proven hilarious over the years. Scrabble champion Will Anderson mercilessly made fun of ChatGPT for repeatedly suggesting playing the non-word OLEICAT. But even though now a lot of people seem to use AI, large language models, and chatbots interchangeably, AI covers a lot of things and I think it’s pretty safe to say that all the Scrabble computing engines that have been used from Maven to Quackle to Macondo count as AIs as well, so AI clearly can play Scrabble and often better than humans. I know the optimal strategy is to think like them. I know I’d probably be capable of it. I just don’t think I’d enjoy it.

I find myself disagreeing with my fellow sports analysts quite often. Not only do I kind of agree that analytics has made sports more boring (I think the real reason a lot of people prefer college basketball to the NBA, college football to the NFL, and the NASCAR O’Reilly Auto Parts Series to the Cup Series is because the teams aren’t as good so their play is less optimized resulting in more mistakes, which can increase the excitement), but I also find myself disagreeing with all these guys especially when it comes to Hall of Fame inductions. <robotvoice>Beep beep boop boop. No players should be inducted into any Hall of Fame unless they are in the top 1.364% of all players to participate in their given league.</robotvoice> The amount of times I have seen my fellow stat bros use phrases like “Hall of Very Good” has constantly made me cringe. If we award too many players, it’ll diminish the honor! Well, not only were almost all Halls of Fame exceptionally permissive in their early years before becoming restrictive later on, but I personally think it’s fun to see people awarded for things. These “Hall of Very Good” arguments remind me very much of all the boomers who complain about us goshdarn millennials and all the participation trophies they awarded us, and I feel so much backlash towards conservatives whining about participation trophies that I’ve got to say at this point, bring them on! More participation trophies! As our world gets darker, we need more things to celebrate. Anything that brings people together should be supported in our atomized time.

I think my biggest gripe is the effects that the quantification and gamification of everything have had on our society. When the world runs on zeroes and ones, that increases the role of money over everything in our lives. Money by definition is easily quantifiable when most of the other things we care about in life are not (love, family, friendship, loyalty, bravery, health, community, companionship, creativity). Any time you try to quantify those things, you tend to fuck them up. Instead of love, we have OnlyFans. Instead of family, we have BetterHelp. Instead of friendship, we have parasocial relationships with streamers. Instead of bravery, we have online gaming. Instead of health, we have health monitoring apps that spy on you. Instead of community, we have social media apps that incentivize hatred because strong emotions increase the likelihood that you’ll become addicted to their platforms. Instead of creativity, we have chatbots.

In my old age, I’ve realized the best things in life aren’t quantifiable and so I’m craving competitive pursuits less and less (yeah, my LearnedLeague entry is a weird exception to that, but I’ll probably quit that after this year is up). Does that mean I’m not enjoying my racing research? No, I am, and it’s one of the only things I’m enjoying in recent years, but I honestly feel I enjoy my roles as a historian and an archivist more than my role as an analyst. I don’t think there’s anything necessarily wrong with being a sports analyst per se and creating new ideas. This does fuel my creative juices in a way I never felt Scrabble did and that’s why I enjoy it more. I also enjoyed writing Nerds per Minute more than I ever actually enjoyed the practice of competitive typing for the same reasons. In a world of AI, creativity (and most of the other items I listed above) are some of the few things it can’t truly replicate, so I’m really valuing them more.

The problem I guess is really with commodifying your work and selling it to multinational organizations like sports teams. I mean what has been the effect of sabermetrics? Mostly, it’s just to find better ways to exploit labor and reduce labor costs. That shouldn’t be a good thing! I realize it can be hard for a lot of precariously employed gig workers/freelancers/writers (which is much of the media now) to feel sympathy for much richer players getting exploited, but the fact is they are and we still see this as a good thing. And the rush to maximize profits or winning chances decreases almost any kind of loyalty. Take Roger Penske’s IndyCar team. He has seemed to have the knack for just releasing drivers at precisely the right moment when they are out of gas: from a business perspective, dropping Ryan Briscoe in 2012 and then eventually replacing him with Juan Pablo Montoya in 2014 was the right move, as was replacing Montoya with Josef Newgarden in 2017, dropping Hélio Castroneves in 2017, replacing Simon Pagenaud with Scott McLaughlin in 2021 (albeit maybe a year too late), and it looks like even replacing Will Power with David Malukas was since Malukas has been a lot faster on road courses than I expected him to be. Having said that, a lot of these sorts of moves can be wrong from a moral perspective (and I’m saying this even though Malukas is my favorite IndyCar driver). Again, it can be hard to sympathize if you’re a broke writer comparing yourself to millionaire athletes, but I personally would rather live in a high-trust society where employers and employees are working together and trust each other rather than being out to get each other. Let’s end at-will employment, please.

I admit it is weird for someone who is doing what I’m doing to admit the best things in life aren’t measurable, but I do think it is correct. I remember right before I quit Scrabble, I had a back-and-forth with Kenji and he told me that I could have become a very good Scrabble player but it wasn’t the right thing for me as a person. He was right and if I ever come back to Scrabble, it will only be for social reasons. I’m desperately lonely and there is nothing I am craving more than real world community. I still don’t feel I’ve found it yet…

RICK KELLY………………….AUSTRALIA

Born: January 17, 1983

Best year: 2011
Best drive: 2004 Eastern Creek Race #1

Although Kelly was one of the most infamous, controversial, and least deserving champions in any major league series, he nonetheless was a reliably consistent Supercars performer for two decades, earning 13 wins including two Bathurst 1000 wins. The younger brother and longtime teammate of 19-time winner Todd Kelly, Rick became the first driver to win 12 races in a season in the Australian Drivers’ Championship (Australia’s top open wheel title) in 2001 before landing his first V8 Supercars ride in 2002 for Holden.

In 2003, Kelly became the youngest Bathurst 1000 winner in a shared drive with his teammate Greg Murphy, who otherwise significantly outperformed him. He achieved this despite the fact that his car owner Tom Walkinshaw ran out of money when his Arrows F1 team collapsed and had to sell off the operation. Holden initially bought it but quickly sold it to Rick and Todd’s parents, John and Margaret Kelly, because manufacturers were forbidden to own Supercars teams. Rick overtook Murphy in 2004, ranking second in laps led and tied for second with 4 wins, including an electrifying drive from 17th at Eastern Creek Raceway and another Bathurst 1000 win.

Garth Tander replaced Murphy in 2005 and typically had the measure of Kelly, but Rick backed into the 2006 title under controversial circumstances despite winning only one race to Tander’s seven and Craig Lowndes’s five. Entering the finale tied with Lowndes on points, Rick bumped Lowndes into Todd and wrecked him. Although he was penalized, Lowndes’s car was damaged enough that Rick still won the title with an 18th-place finish. Rick never really lived that down, but remained a reliable performer, finishing 8th or better in points nine straight seasons from 2003-2011. In 2009, John and Margaret Kelly severed their previous relationship with Walkinshaw and formed their next team Kelly Racing. Their equipment was a lot shakier and Rick only had one winning season from 2009-2017, his three-win 2011. I regard this season as his best since he outperformed Murphy significantly worse that year than in 2004. He was usually the team leader, as he posted winning records against all his regular teammates from 2007-2018 before retiring in 2020. The Kelly brothers, who had taken over their parents’ team in the intervening period, eventually sold it to Stephen Grove in 2022.

Although Rick won fewer races than Todd, I have to say Rick was better as indicated by his 163-89 teammate record and .164 to .048 advantage in my model. He never really achieved the level of greatness that many people might’ve expected considering his initial trajectory, but he remained reliably consistent for a very long time. Since Kelly Racing was one of the few four-car teams in this era, Rick has more teammate comparisons than any other driver in the history of my touring car model, ensuring a high degree of accuracy here. He does indeed trail most 21st century Supercars champions in my model and he had no business winning a championship, but he’s underrated regardless.

Touring car model: #270 of 1676 (.164)

Teammate head-to-heads: 800-476 (1-0 vs. Craig Baird, 18-5 vs. Jason Bargwanna, 8-3 vs. Tim Blanchard, 1-21 vs. Jason Bright, 6-1 vs. Alex Buncombe, 1-2 vs. Matt Campbell, 101-87 vs. Michael Caruso, 2-0 vs. Ben Collins, 1-0 vs. Tony D’Alberto, 65-14 vs. Simona de Silvestro, 1-0 vs. Jesse Dixon, 1-0 vs. Scott Dixon, 6-4 vs. Taz Douglas, 27-2 vs. Paul Dumbrell, 12-8 vs. Dean Fiore, 1-1 vs. Bryce Fullwood, 5-0 vs. Daniel Gaunt, 0-2 vs. Oliver Gavin, 43-39 vs. Andre Heimgartner, 18-8 vs. Garry Jacobson, 1-1 vs. Owen Kelly, 163-89 vs. Todd Kelly, 2-1 vs. Jack Le Brocq, 1-0 vs. Tim Leahey, 1-0 vs. Cameron McLean, 13-0 vs. Mark McNally, 0-1 vs. Tomas Mezera, 58-47 vs. James Moffat, 44-41 vs. Greg Murphy, 0-1 vs. Dylan O’Keeffe, 1-0 vs. Steve Owen, 20-2 vs. Jack Perkins, 2-0 vs. Nathan Pretty, 1-0 vs. Andy Priaulx, 20-4 vs. Karl Reindler, 25-4 vs. David Reynolds, 25-1 vs. Tony Ricciardello, 0-1 vs. Jim Richards, 2-1 vs. Aaren Russell, 9-2 vs. David Russell, 4-2 vs. Alex Rullo, 1-0 vs. Glenn Seton, 1-1 vs. Allan Simonsen, 1-19 vs. Mark Skaife, 3-1 vs. Alex Tagliani, 35-56 vs. Garth Tander, 1-0 vs. Anthony Tratt, 4-1 vs. Jacques Villeneuve, 1-0 vs. Cam Waters, 2-0 vs. Richard Westbrook, 2-0 vs. Justin Wilson, 39-3 vs. Dale Wood)

Year-by-year: 2001: C-, 2003: C, 2004: E-, 2005: C-, 2006: C+, 2007: C, 2008: E-, 2009: C+, 2010: C+, 2011: E-, 2012: C, 2013: C-, 2015: C, 2017: C-, 2018: C+