Transcript

Trial and Error - Actually What is Success | Amy Hoover | TEDxNJIT

[Applause] so we often like to describe things with numbers because we inherently see them with that's some sort of objective in that they are objective so for example there's also the societal drive to try and tackle hard problems with numbers because again numbers themselves indicate some sort of objectivity however the there's also the societal societal trend we'll put it to quantify people and their success so for example you have school teachers and they are evaluated based on the standardized test scores of their students you have police officers who are evaluated based on the number of arrests that they make and perhaps some researchers would know this but but as researchers we are evaluated on the number of citations we have and the money that we are able to bring in to the university but these are hard problems and by while we make it easier to be objective by looking at the numbers how we're quantifying success of these different professions and people we're ignoring the fact that in the in the first place that that these are hard problems to solve and likely are not able to really be distilled in this way oops so let's take let's go back one but let's take this story then to sort as a sort of proof by example of two professors you have professor a and you have professor K so professor a has grown up in this lovely sunny town in Florida with middle-class atomic family a house with a quite large garden as you can see from the picture and a white picket fence it's Florida so we know there's a pool even if even if we can't see it and she as she graduated as valedictorian of her class in high school and went on to have a fully funded college experience during that time she published her first paper as an undergraduate which one of the best paper award and continued on from there to study with the leading researcher and the field that she had chosen she was funded through her extra schooling with a prestigious National Science Foundation graduate research fellowship award it went on to become a professor at a and r1 institution which in the United States is a designation for the top four a top research university in the United States all right so by all accounts it seems like professor a has been successful and then we have professor K so with these illustrations perhaps you can see where this is going professor K was pressured to study something that she was not interested in in particular it had to do with medicine and her parents wanted her to be a doctor and hating the classes that she was enrolled in professor K turned to classic sorts of video games to find meaning in her life she struggled with the demands of the classes she didn't enjoy and considered dropping out entirely from the program she changed her major four times and finally after this fourth attempt she was able to graduate at the end she found a ph.d program that would accept her and at the end of the first semester after having struggled throughout undergraduate because it took her six years to graduate at the end of the first semester of her PhD program her father was killed in tragic car accident and before her mother could see / future professor kate graduate her mother was also killed in a separate but just as tragic car accident however professor k is also a professor at an r1 university and so if you're going to put a number on me as both professor a and professor k for this proof by example you'll see that if you evaluate me as a researcher these two people will be the same but these are very different narratives to tell one is a story of privilege that preordained success in some ways where this this road to really the road to success has has been paved by privilege in the other is a more of a struggle of trial and error and the desire to succeed and overcome different obstacles in her life to be resilient but in particular with professor k is this notion of trial and error and the narrative of professor k that is affecting not only the narrative of my life but also in my research and so i have two pieces of advice that may be helped that i've discovered through my life and through my research that may be helpful when you yourselves are faced with some adversity the first is to consider the significance of a single failure because life is noisy there is a lot of randomness that the one failure does not mean that you are a failure it's not scientifically significant and it shouldn't define you and so in my research this is an image of a game called hearthstone which is a card game i try in my research which work in AI in a type of AI called evolutionary computation and reinforcement learning I tried to create better algorithms to solve bigger and harder problems well with these decks we are trying to we're with this game hearthstone we're trying to find from a big space of possible cards what are the best ways to combine them and this is relevant because these decks need to be evaluated so if I look at the value of one deck I may look at what does it win or not and how many turns does it take to win so I play that card deck against another player once and I see that in this case it takes 10 turns but this deck is good it's valuable and it won so I should keep it but that same deck when you are playing it against the same opponent and player you may find the it wins and what are it losses in one turn which is a significant loss and so what I need to do in my research is then evaluate these decks a large number of times close to 400 to decide whether or not this difference is significant there's a before we get to this next slide there's another example of famous papers written by researchers that at the time were not famous that were rejected from the conferences and journals to which they eventually or to which they had to applied and these these papers are things that you may read about in your biology textbooks or your computer science textbooks and so again with this narrative of a single failure is is not necessarily indicative of a problem with you that it may behoove you to keep trying something else that I see and my research and that I've seen in my life is that there can be many definitions of success and so I work with a type of algorithm that doesn't just look at one type of success but looks at success over a span of dimensions so a way of large variety of different ways of defining what success means and so in this same game I'll go back to this for a second in this same game hearthstone we talked about the deck evolution but we also look at game play strategy and what that means is you could play aggressively or you could play much more controlled and we find in in the experiments that we've been running that instead of if we're not searching for the best aggressive strategy and we're not searching for the best control strategy but instead searching along a variety of different measures of success that we find strategies that would that perform better than either of these aggressive or control strategies would have alone and so from my experience as Professor a and Professor K and my life I can say that and from my research findings that trial and error is important and I would encourage you to continue to explore different avenues of what it means for you to personally be successful and to be prepared for some singular failures and not take it as a as an indication of your value that success takes many attempts sometimes and it's taking to really define yourself in order to be successful with your research results but it can take many attempts and there are many different ways to evaluate success so perhaps we're back to the original examples of teachers police officers and we'll we've researchers a sigh that we should likely be having a bigger conversation about what it means for all of us to be in all of these professions to be evaluated is it bright or fair to have these metrics for those individuals and so really what I'm trying to get at is that through my life and through my work I found that you really don't know the value of a given strategy until you've evaluated it many many times thank you very much [Applause]