Alliance Captain versus last pick in Alliance Selection

[Joe G.]

For me, 1A and 2A, without a doubt. I have had a few experiences being dragged to high placement as the last pick (and one being carried by excellent partners as an unworthy captain), with lousy, dysfunctional robots. They weren't fun, but they developed a very strong ability to gauge success through metrics other than final placement. To me, placing as a captain or early pick that doesn't make it far almost always represents a greater accomplishment for a team than a 3rd robot, even a well-done, role-player 3rd robot.

The "B" options are interesting because, while mid-range alliances that get eliminated quickly tend to be forgotten by the greater community, everyone in the community can point to several "golden examples" of picture-perfect 3rd robots, which were extremely simple to build, did their job perfectly, and had a major impact on their alliances. 1503 in 2011. 4334 in 2012. 148 in 2008 (calling this one "simple," or a good example of low resource design kinda baffles me, but a lot of people do so and it works with my point, so I'll include it). That pure ramp bot that won your local regional in 2007. And so on. And it's very easy, and in many ways completely correct, for a team to look at these success stories, decide "we should do that!", and build a robot which explicitly targets this 3rd robot position. Especially if part of what a team is after is greater recognition in the community.

The problem with this mentality is twofold. First, contrary to popular belief, it's actually really hard, and I would argue not always fully possible, to be an ideal 3rd robot through designing to be a role player, or at least to spend a season with the intent of being an ideal 3rd robot. Building a sophisticated robot capable of being a contender is hard in some very real, obvious ways. It takes a lot of engineering skill to design effectively. It takes a lot of money to put together. It takes a lot of time to assemble it, program it, and practice with it enough to get the most out of it. Building a robot like 4334, on the other hand, is hard in a very different way. It's hard to do in-depth enough game analysis to be reasonably confident in your decision to depart from the "expected" strategy. It's hard to have the guts to intentionally design a robot that will probably not win a regional, and depend on other means to get to the world championship to play the role it was meant for. It's hard to convince stubborn people to abandon the "exciting" parts of the game. It's hard to judge where to draw the line between strategic flexibility, and unneeded features which take away from the core objective. It's hard to define simplicity, to judge exactly how simple a given mechanism will be, and determine whether or not a means of adding percieved simplicity is going to make your life more difficult in the long run, or paint yourself into a corner (My team messed up bigtime on this one this year). It's hard to design a robot, even a very simple one, that radically departs from the "stock" design from the past game this year's game follows closely, from Ri3D and other prototypes emerging, especially when success at 2B essentially depends on near-flawless execution. It's hard to convince teams that your little robot is the best possible partner for them, especially if your chosen strategy really doesn't play out very much during qualifying matches.

And it's very hard to pick out the winning, genuinely useful support roles from those that just don't work out. Some examples of robots I have seen from each of the past few years, where the team behind them probably truly believed they were building "the next 4334" at the time:

• 2015: Pure canburgling robots, with well made, robust, well-engineered systems, which were too integral to the design to be upgraded enough to be competitive with elite or even semi-elite "swinging post" style canburglers. 2015 was an especially bad year for this, as at high levels of play, 3rd partners were either entirely cheesecaked creations with none of the initial design goals/intent present in the finished product, or 2nd tier human loaders with the exact same strategic goals and design elements as their slightly faster captains. Rigid specialization just didn't work out for anybody this year, and it won't necessarily work out in future years. This is another wrinkle in the difficulty of the design process for one of these robots -- teams may single-mindedly seek out a strategy that fits this mold, when an effective one really doesn't entirely exist.
• 2014: Pure goalie robots designed to man the goalie zone in teleop, with either no ball manipulation at all, or only as a total afterthought. Robots centered around catching.
• 2013: In most cases, otherwise immobile pure 30 point climbers. This one doesn't quite fit as well, since I think most of the teams who did this primarily under performed due to underestimating the difficulty of the 30 point climb, instead of strategic error. I also saw a "feeder station" robot, designed to emulate the human loader station and ferry disks down-field to their partners, which ended up just slowing things down. It didn't work out well.
• 2012: This year, most of the side/specialist objectives were actually highly worthwhile if performed with high quality execution, one of the reasons that 2012's probably my favorite game of all time. Still, I ended up seeing a lot of robots that perceived certain specialty tasks (bridge tipping and facilitating triple balances, mostly), as being much more difficult and involved than they were, and developed huge systems that encompassed their entire robot to achieve these things with authority, when in the end, almost negligible additions to otherwise fully featured robots like passive wedges or stingers ended up outperforming these dedicated designs. This is yet another area where teams going for this kind of strategy need to be extremely careful: If you zero in on a task and devote your robot to it, it's very easy to allow the design freedoms that this gives you to cause your design to spiral out of control, or skip straight past the easiest solutions because of the perceived difficulty of the task, when simple passive additions can often be the very fastest or most reliable way of accomplishing a task. Again, look at teams who centered their entire robots around canburgling, vs. those who used dead-simple auxiliary systems. Which ones were winning races?

Ultimately, while I think that it's extremely important for teams to build within their means and be realistic with their design goals, designing explicitly to be a specialty-role 3rd robot can actually be a much bigger risk with less payoff than building a mid-level robot with very conventional objectives. I admire the teams that do this and succeed greatly, but I also think that there's some of the "monkeys at typewriters" effect in play: there's an element of luck in identifying subtle, alternative strategies or nuanced specialty roles in a game that has never been played before, and with enough FIRST teams working away at the problem, there will be some bad solutions, some good solutions, and the occasional great solution. Of course, the more skilled, insightful, and dedicated you are, and the more you work at it, the greater your chance of being one of the "great solutions." But I think that it's misleading to describe this approach as safe, or low risk.

Additionally, back to the question of "which type of robot represents "success," or team growth, better?", I'd point to the history of the classic examples of B-type teams in the years after their B-type year. I can't think of any teams that consistently target, and nail, this type of strategy and role year after year. It seems like most of them either float around this approach for years, and are very hit-and-miss during this time, or move on to more conventional designs, and may ultimately reach positions as alliance captains, but not necessarily immediately. 4334 built a full court shooter which missed elims at their first regional, then a pretty conventional 2014 robot (albiet with a revolutionary stupid-simple strategic innovation which permanently altered the game dynamics). 148 built upon their 2008 season to become one of FIRST's biggest powerhouses, but nobody would call their designs these days "simple." And so on. It's clear that the teams who have walked this path see growth beyond being the 3rd robot as a valuable thing for their program, even if there's a period where they don't get quite as many blue banners.