Scouting, is it important?

[EricH]

Any robot can be beaten, with the right strategy.

The problem is, you need to know that strategy.

To remove the scouting from the equation removes any chance of your knowing that strategy.

I'm going to go back to 2005. With the #10 pick, 330 picked team 503. We would not have known that they were a diamond in the rough if we had not been scouting. We were very surprised that they were still available. A few hours later, and teams 330, 67, and 503 were hoisting some largish trophies and wearing some nice, shiny new medals after winning the Championship. Without scouting, there is a very good chance that we would have been knocked out in the division eliminations.

Eliminate your scouting--and have fun being eliminated. Keep your scouting, and improve it.

When I'm scouting, I couldn't care less about the technical stuff like # of wheels, arm/lift. What I care about if I'm scouting is: What is the robot doing in automode? What is the robot doing in teleop? What is the robot doing in the endgame? How well is it doing it? Does it break down often? And finally, How well does their team interact with our team? The last item is the only one that can't really be done objectively.

I'm also going to add on to what Joe was saying. We've done pen/pencil and paper for a while; the team has recently started doing other methods. (Wiimote, or something like that, IIRC). The data is input into a computer, and the pit scouting stuff is in a different place. Friday night strategy meeting goes something like this: "How do we want to play our matches tomorrow?" "Well, teams X, Y, Z have this, that, and the other, so this might be good." "OK, now, who do we want to pick/be picked by?" *hour-long debate culminating in "Keep an eye on these X teams, and if you see anybody else, let us know. These Y teams, we don't want to pair with, but don't tell them that."*

In 2005, we had a simple system: How many tetras did a team score, and where? 2006, with game objects flying everywhere, we used a "on a scale of 1-3, how good are they at X" along with "Where do they go in automode?" 2007 returned to a how many/where scouting method, which would also work for 2010 (and 2008).

[rcmolloy]

Originally Posted by EricH

Any robot can be beaten, with the right strategy.

The problem is, you need to know that strategy.

To remove the scouting from the equation removes any chance of your knowing that strategy.

I'm going to go back to 2005. With the #10 pick, 330 picked team 503. We would not have known that they were a diamond in the rough if we had not been scouting. We were very surprised that they were still available. A few hours later, and teams 330, 67, and 503 were hoisting some largish trophies and wearing some nice, shiny new medals after winning the Championship. Without scouting, there is a very good chance that we would have been knocked out in the division eliminations.

Eliminate your scouting--and have fun being eliminated. Keep your scouting, and improve it.

When I'm scouting, I couldn't care less about the technical stuff like # of wheels, arm/lift. What I care about if I'm scouting is: What is the robot doing in automode? What is the robot doing in teleop? What is the robot doing in the endgame? How well is it doing it? Does it break down often? And finally, How well does their team interact with our team? The last item is the only one that can't really be done objectively.

I'm also going to add on to what Joe was saying. We've done pen/pencil and paper for a while; the team has recently started doing other methods. (Wiimote, or something like that, IIRC). The data is input into a computer, and the pit scouting stuff is in a different place. Friday night strategy meeting goes something like this: "How do we want to play our matches tomorrow?" "Well, teams X, Y, Z have this, that, and the other, so this might be good." "OK, now, who do we want to pick/be picked by?" *hour-long debate culminating in "Keep an eye on these X teams, and if you see anybody else, let us know. These Y teams, we don't want to pair with, but don't tell them that."*

In 2005, we had a simple system: How many tetras did a team score, and where? 2006, with game objects flying everywhere, we used a "on a scale of 1-3, how good are they at X" along with "Where do they go in automode?" 2007 returned to a how many/where scouting method, which would also work for 2010 (and 2008).

Extremely well put Eric. I personally believe that scouting is about 70% of you winning the match. The other 30 is from performance and human error. This year tested the ways of scouting much more because of the various ways the robots could play. I believe it's like not just an advantage over other teams but also the best possible way to have your team succeed.

[carbuff2228]

hi i am 2228 scouting and strategy lead for the four years ive been on the team. The person who started this thread is the coach for my team as well. I see what he means by how scouting is useless. I have to slightly agree with him just do to not enough commiitment from people match scouting for our team. I think those of us who love scouting like me understand how criticial and important it is. I mean people tell me a team number and i know the name andf where there from and i am extremely dedicated. But i think you need just a few good eyes match scouting and do some pit scouting and youll do fine. Our scouting this year in North Carolina helped us pick two good teams 401 hokie guard and 23614 mars and we did decent in the QF. But i think youll always need scouting and is the most important thing to do at regionals

[JesseK]

At some point, you will inevitably lose a match you're supposed to win. With good scouting, you may win a match (or many matches) you're supposed to lose. Of course this assumes all else equal.

If no one on the team is into scouting or into the 2nd-regression statistical analyses that may be significantly educational, then perhaps its best you do away with scouting. In my experience the lack of scouting generally points to a lack of team cohesion and the commitment to win even with a sub-par robot, thus there are bigger issues to address.

[BrendanB]

Originally Posted by rcmolloy

Extremely well put Eric. I personally believe that scouting is about 70% of you winning the match. The other 30 is from performance and human error. This year tested the ways of scouting much more because of the various ways the robots could play. I believe it's like not just an advantage over other teams but also the best possible way to have your team succeed.

Extremely well put here! Two extremely good robots cannot just walk onto a field and play to win. For example, one very good robot who plays strictly near cannot pick another strictly near robot who is also extremely good just because they score tons of balls, you will just end up with a really bad mid zone robot with a good near zone robot getting torn apart.

One thing our team added this year was time in each zone to determine where a robot likes to play.

[Peter Matteson]

Originally Posted by Joe Ross

What you've described is also scouting. We try to do both objective scouting (like the paper sheets you describe) as well as subjective scouting where a few smart people try to figure out the teams strategy and tendencies. We scout both for strategy for the next match, and for choosing who to pick. In both cases, we use the objective data as the starting point, and then modify our plans based on the subjective data.

Just like our robot, we improve the scouting throughout the season. We usually find that at the first regional we either collect too much objective data, or not the right types of objective data.

Joe got what I was getting ready to type spot on.

We learn what to look for by watching a week one regional and make our first rev of our scoutning plan. This is heavily influenced by our stratagy team deciding what we need to know to pick and what we need to know to win the match. If we learn after one event we didn't collect the right info we update our plan and move on.

All the info on each team gets filed into "the bible" after each match we watch. Our qualitative scouters work with the drive team to create stratagies that give us the best opportunity to win the match.

It's important not to just stick people as scouts, they need to want to do it and understand what it does for the team. Our group of 7-10 scouters always has a leader who wants to be there and keep things going because they know our method of scouting has done well for us over the years. If the scouts don't buy into the system and the plan then you get garbage output. Make sure that they feel they are contributing and not being wharehoused for the event.

[smurfgirl]

Originally Posted by Peter Matteson

All the info on each team gets filed into "the bible" after each match we watch. Our qualitative scouters work with the drive team to create strategies that give us the best opportunity to win the match.

...they know our method of scouting has done well for us over the years.

This is one of the biggest understatements I have ever heard. 177's scouting system is both fabulous and incredibly comprehensive. And if you look at their track record... well, their consecutive appearances on Einstein for the past five years and their two Championship wins are definitely heavily correlated to their excellence at scouting.

This is the part where I want to jump in with one of my favorite FIRST stories about how 1124 learned everything we know about scouting and countless other lessons from Team 177 back in 2006. We didn't really understand the purpose of scouting then, nor did we do much of it. We had some bad luck in our matches, and were ranked at 40/40 teams for a while on Saturday morning, but we moved up a few places to 34/40 after our final qualification match. We figured that with a number that low, there was no way anyone would ever pick us, and it was over for that competition. However, this was not so- after pairing up with Team 176, Team 177 invited us to join their alliance. And boy, had they done their scouting homework. In strategy discussions among the alliance, they shared information about other teams that blew our minds, because we were naive and understood virtually nothing about scouting. They knew exactly how to out-strategize each opposing alliance. To make a long story short, we played a lot of matches with the alliance because under their scouting-informed direction we kept winning them, and they opened up our eyes to the value of scouting, how to scout, and lots of other wonderful lessons that helped our team improve. (It was a bit more complicated for me; it was a magical experience for me because it was my first regional and a lot of things started to fall into place because of them... but the bottom line is still that scouting is important, and team 177 knows what they're talking about when it comes to scouting.)

[SammyKay]

Having been on drive team for the first two years of my time on team 316 and then being back up drive team this past year, I have to say that scouting is a HUGE part of succeeding in competitions. If you have a strong scouting team, then each match becomes slightly easier for you. Before each of our matches this year, we had one mentor who would come down and talk to our driver and tell her what our opponents had done in their last two or three matches. She would also tell our driver whether our alliance partners had broken down during a match or not so we could prepare ourselves for a loss of an alliance partner if it happened again. Even though many of us don't like to do match scouting, it is extremely useful for our team, especially if we modify our scouting techniques to fit the specific competition.

[rcmolloy]

Originally Posted by SammyKay

Having been on drive team for the first two years of my time on team 316 and then being back up drive team this past year, I have to say that scouting is a HUGE part of succeeding in competitions. If you have a strong scouting team, then each match becomes slightly easier for you. Before each of our matches this year, we had one mentor who would come down and talk to our driver and tell her what our opponents had done in their last two or three matches. She would also tell our driver whether our alliance partners had broken down during a match or not so we could prepare ourselves for a loss of an alliance partner if it happened again. Even though many of us don't like to do match scouting, it is extremely useful for our team, especially if we modify our scouting techniques to fit the specific competition.

Sammy is exactly right everyone. Not going to lie, the matches where we were hectic and running around trying to talk to teams to see what they knew really hurt us. Although, there were people from the stands giving us minor hints about the robot. (Saying if it was good or bad...not really helpful haha) BUT, in those matches, I had found it much easier to talk to other teams about strategy.

So I am guessing this topic has a lock for scouting being EXTREMELY important??

[Tetraman]

Will scouting always be some kind of "secret weapon" that teams must do on their own?

[EricH]

Originally Posted by Tetraman

Will scouting always be some kind of "secret weapon" that teams must do on their own?

Each team is looking for something different. As such, they will want different data about the same robot.

That said, if two or three or more teams were to collaborate on scouting, nobody would look askance at it. However, the data would need to be shared equally, and the work done roughly evenly, if possible. (Or, a pre-arranged agreement made, something along the lines of, we'll provide X students to help you gather data, in exchange for Y data set that you're already collecting.)

By the way, I do know that in the past, 330 has allowed teams access to what our scouting data showed about their performance. 1114 does somewhat of the same thing by releasing their scouting report before the Championships.

[Vikesrock]

Originally Posted by Tetraman

Will scouting always be some kind of "secret weapon" that teams must do on their own?

There are is a fantastic presentation on scouting/strategy written by Karthik that can be found on the 1114 website. I have a presentation that I gave at MN Splash that was based off of Karthik's and can be found at mnfirstregional.org

In 2009 Team 2175 collaborated on scouting at the Northstar regional with Team 171. In 2010 we again worked with 171, but a number of other teams joined in, by the end of the event the data was being shared with 10+ teams. There were definitely both pros and cons to doing things that way.

[angelawence11]

Scouting is incredibly important. I mean, the information isnt always used, but we have been in numerous situations in which we have consulted with our lead scouts for advise on teams to pick. I mean, sure the first place team is usually a good team to pick. But in some games even the best robots can end up with a bad seeding. In lunacy, for instance, the game was very different, and sometimes the best robots wouldnt even be picking. Or the Uberbots this year. they had a fantastics robot, but sometimes their seeding wouldnt be as high because they would play defense sometimes. They usually ended up doing great, but the scores of a match dont always reflect the bots capabilities.

[Chris is me]

Originally Posted by Tetraman

Will scouting always be some kind of "secret weapon" that teams must do on their own?

While scouting is the most important thing my team does at competition, I'm not opposed to doing it with others. At WPI, team 20 was kind enough to lend us their data, while at CT we used our own with the help of sgreco on Saturday.

I do think it is 1000% understandable for any team to keep their pick list completely secret. 2791 is the kind of team that shares anything and everything, yet there are many reasons we do not make our pick list with others nor do we share it except with our first pick.

However, with regards to data collection and the way we organize information on teams, I'd be happy to share any of that with anyone. I actually have been looking into doing a pre-Kickoff presentation on the subject, since not everyone can see Karthik's wonderful presentation on scouting.

[JamesBrown]

Originally Posted by EricH

That said, if two or three or more teams were to collaborate on scouting, nobody would look askance at it. However, the data would need to be shared equally, and the work done roughly evenly, if possible. (Or, a pre-arranged agreement made, something along the lines of, we'll provide X students to help you gather data, in exchange for Y data set that you're already collecting.)

This is not necessarily the case. 358 has twice given 3280 their scouting data and access to their scouts because we didn't have enough people to effectively scout on our own. In both of these situations 358 was seeded below us. In one of the two 358 was seeded below us but was also with in the top 8, meaning we were picking the robots that they wanted before they had the opportunity.

From a match prep perspective I don't think I have seen a better system than 1519. Every match I go into I feel that I have a good idea of what the other bots will do. 1519 has concrete data that shows what they will do.