Discussion: Score-O Tactics
in: Orienteering; General
Aug 16, 2020 2:08 AM
#
I am trying to come up with guidelines for tactics in Score-O route planning for competitors. I will probably incorporate these into a document/presentation that will be used for instructing purposes.
Generally, Score-O is a discrete case of the Travelling Salesman Problem (TSP). There are a few algorithms that are used in computer science to tackle that through a set of calculations and simulations however in field conditions all you have is your good judgment and past experience.
Of course, there are different formats to Score-O: events in which all controls have the same value, events that controls differ in value, different penalties for being late, etc. Still, I’m trying to come up with set of heuristics/tips/practicalities.
Any insights would work! Feel free to share anything from practical advices (e.g. use a string to estimate distance) all the way to taking a scientific approach (e.g. instantiation of known algorithms).
Examples:
- Bring a string, that would resemble the distance that you expect to run given your current shape in the given time, and curve it along the route that you plan on taking
- Designing your route as a loop is more efficient than having a route with one or more intersections (potential paths that cross one another) - minimize the number of intersections
- The 2-OPT technique (a known method in computer science to solve TSP) involves iteratively removing two edges (legs) and replacing these with two different edges that reconnect the fragments created by edge removal into a new and shorter tour – use that technique to optimize your planned route
- Save some controls that are relatively close to the finish to the last few minutes – these can be collected towards the end or skipped if need be
- Identify areas “saturated” with high value controls (in case that controls differ in value) or areas condensed with controls (in case that all controls are equal in value) – planning your route to cross through these areas will increase your return for the effort of getting there
- Shorter legs are preferred over longer legs
- Plan a course that would allow you to punch controls all throughout the race - on the way out and on the way back
Thanks everyone!
Generally, Score-O is a discrete case of the Travelling Salesman Problem (TSP). There are a few algorithms that are used in computer science to tackle that through a set of calculations and simulations however in field conditions all you have is your good judgment and past experience.
Of course, there are different formats to Score-O: events in which all controls have the same value, events that controls differ in value, different penalties for being late, etc. Still, I’m trying to come up with set of heuristics/tips/practicalities.
Any insights would work! Feel free to share anything from practical advices (e.g. use a string to estimate distance) all the way to taking a scientific approach (e.g. instantiation of known algorithms).
Examples:
- Bring a string, that would resemble the distance that you expect to run given your current shape in the given time, and curve it along the route that you plan on taking
- Designing your route as a loop is more efficient than having a route with one or more intersections (potential paths that cross one another) - minimize the number of intersections
- The 2-OPT technique (a known method in computer science to solve TSP) involves iteratively removing two edges (legs) and replacing these with two different edges that reconnect the fragments created by edge removal into a new and shorter tour – use that technique to optimize your planned route
- Save some controls that are relatively close to the finish to the last few minutes – these can be collected towards the end or skipped if need be
- Identify areas “saturated” with high value controls (in case that controls differ in value) or areas condensed with controls (in case that all controls are equal in value) – planning your route to cross through these areas will increase your return for the effort of getting there
- Shorter legs are preferred over longer legs
- Plan a course that would allow you to punch controls all throughout the race - on the way out and on the way back
Thanks everyone!
Aug 16, 2020 2:58 AM
#
Sprint to the nearest control then figure out the rest along the way
Aug 16, 2020 8:24 AM
#
GJN:
Remember to take into account the various Laws of Score Events, established by scientific research over many years. These include:
the Law of Map Scales: all score events take place across the edge of a distortion in the space-time continuum. Consequently, having run for 45 minutes on a 1:10 000 map, and whilst at the furthest part of the area, you will realise that the map has transmuted to 1:15 000 so there is no way you can get back in time.
the Laws of Easy Controls:
1) Always leave a couple of low-scoring controls to pick up on the way back to the finish
2) The control which you cannot find will always be one of the low-scoring controls you have left to pick up on the way back to the finish.
the Unseen Map Rule: the fastest person away from the mass start, being followed by the greatest number of people, is invariably the one who has not yet looked at his map.
the Law of Diminishing Returns: planning their routes at the start, M45s forget that they are no longer M21s.
the Law of the Second Coming: only as you actually leave the control will you recognise having been there 40 minutes ago.
the Fifty Minute Rule: fifty minutes into the hour, you will simultaneously realise that:
i) you are at least 12 minutes' run away from the finish
ii) you have misread the contours, and that the long down-hill run-in you had left yourself isn't.
the Law of Folded Maps: on rare occasions you will finish two minutes early. It is in those two minutes that you will spot the control you missed out because of the way you had your map folded.
the Law of Map Scales: all score events take place across the edge of a distortion in the space-time continuum. Consequently, having run for 45 minutes on a 1:10 000 map, and whilst at the furthest part of the area, you will realise that the map has transmuted to 1:15 000 so there is no way you can get back in time.
the Laws of Easy Controls:
1) Always leave a couple of low-scoring controls to pick up on the way back to the finish
2) The control which you cannot find will always be one of the low-scoring controls you have left to pick up on the way back to the finish.
the Unseen Map Rule: the fastest person away from the mass start, being followed by the greatest number of people, is invariably the one who has not yet looked at his map.
the Law of Diminishing Returns: planning their routes at the start, M45s forget that they are no longer M21s.
the Law of the Second Coming: only as you actually leave the control will you recognise having been there 40 minutes ago.
the Fifty Minute Rule: fifty minutes into the hour, you will simultaneously realise that:
i) you are at least 12 minutes' run away from the finish
ii) you have misread the contours, and that the long down-hill run-in you had left yourself isn't.
the Law of Folded Maps: on rare occasions you will finish two minutes early. It is in those two minutes that you will spot the control you missed out because of the way you had your map folded.
Aug 16, 2020 8:34 AM
#
Ifor:
Depends how long the event is. If it's a short hour event you can not afford too much thinking time I like to be moving in 10-20 seconds for an hour event. If it's multi hour then I may take a minute or two over the planning but never too long.
Depends what your chances are of getting them all. I typically have a feel about getting them all before hand based on the size of the area, If I don't know the area just a scale and size of the map and the hilliness is a starting point. Once I get the map are the controls all the way out the the corners or a bit more tightly clumped will refine my judgment.
If I really think I can get close to all of them then I find the most awkward isolated control and choose the direction such that it is towards the end of the time so it's there as a starting point for shortening. I have a timer set to beep such that I will know when at half time and use this to do a quick re-assessment as I go.
If I don't think I can get them all as is normally the case now then I look for which section is more isolated and decide to skip this section from the start. I concentrate on getting most of the others in the area I am going to but will avoid anything that is clearly longer than the norm for it's score.
Generally I find there is either something obvious route wise or it's not obvious and there are lots of potential similar route. If the second don't worry just get going on something and don't waste too much time thinking about it at the start. In the none obvious types I find an inner and outer loop often works well. I refine if controls between the loops fit best on the inner or outer as I go. if it's not obvious then typically these are the same controls as you want to be potentially missing if running short on time so put them off until later when time pressure may mean you don't need to go there on either loop.
Most of my score events are either multi hour MTB or 1 hour night score events. For biking I pay more attention to the climb direction as often one way is preferable going up or down. For the night O especially on a technical area I look to identify any really hard looking controls and if they have a clear better direction of attack then I can take that into account deciding the route.
Don't wast too much time looking for a control. If you miss it be prepared to just take the hit and carry on rather than going back and having another go.
For the events I do it's typically worth being slightly late back especially the MTB events where up to 10 minutes late is often OK but no more with the sliding scale of penalties. For the Night events then a few seconds over is normally positive for the overall score. The penalties are at the same rate per minute as collecting all the controls in the time limit.
All to be tested for the first time in 5 months latter today. I don't expect to get anything like all of them and want be looking to push the time limit at all as just keeping moving for the hour in terrain will be pushing my running fitness to the limits. Just hope I have remembered how to navigate.
Depends what your chances are of getting them all. I typically have a feel about getting them all before hand based on the size of the area, If I don't know the area just a scale and size of the map and the hilliness is a starting point. Once I get the map are the controls all the way out the the corners or a bit more tightly clumped will refine my judgment.
If I really think I can get close to all of them then I find the most awkward isolated control and choose the direction such that it is towards the end of the time so it's there as a starting point for shortening. I have a timer set to beep such that I will know when at half time and use this to do a quick re-assessment as I go.
If I don't think I can get them all as is normally the case now then I look for which section is more isolated and decide to skip this section from the start. I concentrate on getting most of the others in the area I am going to but will avoid anything that is clearly longer than the norm for it's score.
Generally I find there is either something obvious route wise or it's not obvious and there are lots of potential similar route. If the second don't worry just get going on something and don't waste too much time thinking about it at the start. In the none obvious types I find an inner and outer loop often works well. I refine if controls between the loops fit best on the inner or outer as I go. if it's not obvious then typically these are the same controls as you want to be potentially missing if running short on time so put them off until later when time pressure may mean you don't need to go there on either loop.
Most of my score events are either multi hour MTB or 1 hour night score events. For biking I pay more attention to the climb direction as often one way is preferable going up or down. For the night O especially on a technical area I look to identify any really hard looking controls and if they have a clear better direction of attack then I can take that into account deciding the route.
Don't wast too much time looking for a control. If you miss it be prepared to just take the hit and carry on rather than going back and having another go.
For the events I do it's typically worth being slightly late back especially the MTB events where up to 10 minutes late is often OK but no more with the sliding scale of penalties. For the Night events then a few seconds over is normally positive for the overall score. The penalties are at the same rate per minute as collecting all the controls in the time limit.
All to be tested for the first time in 5 months latter today. I don't expect to get anything like all of them and want be looking to push the time limit at all as just keeping moving for the hour in terrain will be pushing my running fitness to the limits. Just hope I have remembered how to navigate.
Aug 16, 2020 10:46 AM
#
You should always go for just one more control near the end even if you know you won't make it back in time. That way other people can't tell you at the finish that you had a few minutes/seconds left so should have gotten another one. This also applies even if you've already cleared the whole course.
Aug 16, 2020 12:56 PM
#
- Consider climb: In very hilly areas especially, look for ways to string controls of similar elevation together. (often for a 3 control sequence, the shortest distance is high-low-high, but a slight to moderate addition in distance to sequence them high-high-low can be well worth the extra distance)
- Consider the network of trails and roads. Again, a longer route to a different sequence of controls may have big benefits over cutting straight through terrain.
- Consider attackpoints (generally this applies more to longer duration score events like rogaines and adventure races where the map may lack detail and feature descriptions may be vague, but even on a regular map, modifying your sequence to give easier attack points can sometimes have value).
- Consider the network of trails and roads. Again, a longer route to a different sequence of controls may have big benefits over cutting straight through terrain.
- Consider attackpoints (generally this applies more to longer duration score events like rogaines and adventure races where the map may lack detail and feature descriptions may be vague, but even on a regular map, modifying your sequence to give easier attack points can sometimes have value).
Aug 16, 2020 1:02 PM
#
Before starting out, quickly scan the map for impassable obstacles (especially long ones) and crossing points.
Aug 16, 2020 1:25 PM
#
Bash:
Lots of good suggestions here. In events with a night portion, see if you could follow a main trail/road during some of the hours of darkness, picking up controls on both sides. A dark road run can also be an efficient way to move from one area of the map to another. When I choose a 24-hr route, there is often a part of the route that looks better for the night so I might reverse my direction to get myself there at the right time.
Aug 17, 2020 1:13 AM
#
it all really depends whether you have time to find all controls, or not. If not, the individual points of each control is important to determine a good route.
I find the best way to determine a good route is the "finger test": run you finger over the course, from leg to leg. Try to make obvious connections first - a series of controls which obviously connect, maybe because they're along a path, or along the periphery of the map or other. This will create a series of good course fragments.
Then (always tracing with your finger) try to find good connections betwen those "good" fragments. Sometimes this leads to obviously bad legs, in which case you might have to break up one of the fragments you found first.
Of course you'll want to make good use of elevation already gained - don't lose it for nothing! Good attackpoints, good runnability are other criteria for chosing legs.
When you do this with your finger, you automatically develop a "flow" around the course. Your finger will run smoothly along paths and obvious lines. It will tell you it can't go over this cliff, or swim over the lake - it stops, the flow breaks. Try another path.
Some further thoughts that might help, not strictly mathematically correct, but most of the times true:
- your route should never intersect with itself
- avoid acute angles in your course
- maximize the area your route encloses on the map
- once you have your route, you still have the option of doing it one way round or the other. Check both options before deciding (steepness of climb, attackpoints).
If you won't find all controls, it might be a winning strategy to invest some time and run into an area with high-point controls at the beginning. From there make your way back. Basically, if points are a factor 5 higher, but you only invest 4 times more time to find them (compared to those around the start), then it's already worth it.
I find the best way to determine a good route is the "finger test": run you finger over the course, from leg to leg. Try to make obvious connections first - a series of controls which obviously connect, maybe because they're along a path, or along the periphery of the map or other. This will create a series of good course fragments.
Then (always tracing with your finger) try to find good connections betwen those "good" fragments. Sometimes this leads to obviously bad legs, in which case you might have to break up one of the fragments you found first.
Of course you'll want to make good use of elevation already gained - don't lose it for nothing! Good attackpoints, good runnability are other criteria for chosing legs.
When you do this with your finger, you automatically develop a "flow" around the course. Your finger will run smoothly along paths and obvious lines. It will tell you it can't go over this cliff, or swim over the lake - it stops, the flow breaks. Try another path.
Some further thoughts that might help, not strictly mathematically correct, but most of the times true:
- your route should never intersect with itself
- avoid acute angles in your course
- maximize the area your route encloses on the map
- once you have your route, you still have the option of doing it one way round or the other. Check both options before deciding (steepness of climb, attackpoints).
If you won't find all controls, it might be a winning strategy to invest some time and run into an area with high-point controls at the beginning. From there make your way back. Basically, if points are a factor 5 higher, but you only invest 4 times more time to find them (compared to those around the start), then it's already worth it.
Aug 17, 2020 2:45 AM
#
If the question is how to design a Score-O, the answer should be how to best thwart the plans of the competitors. The designer should attempt to make them second guess their decisions, lure them into traps and spend too much unproductive time. Every competitor is different and every terrain and format is different. This is an art as well as a science. Ample time, creativity and field work should go into designing a high quality Score-O.
Aug 17, 2020 3:48 AM
#
Dont be dogmatic about rules. On rogaines I often use a very acute angle - out and back. Its amazing what a lift you get when you offload your backpack for a while. Mind you I once noticed that my pack seemed to be heavier than it should have; inspection revealed a rock that wasn't there at the start...
Aug 17, 2020 5:54 AM
#
Great discussion so far, a few gems in my opinion (just a few examples):
"Points on the periphery of the map can be connected"
"Make use of elevation - don't lose what you've earned"
"In very hilly areas especially, look for ways to string controls of similar elevation together."
"Quickly scan the map for impassable obstacles"
"Follow a main trail/road during some of the hours of darkness, picking up controls on both sides."
"Avoid acute angles in your course"
"...option of doing it one way round or the other."
"I have a timer set to beep...at half time"
And I love the idea behind"finger test"
A couple of comments that I am not sure that I agree with (disclaimer - this is just my opinion):
"Sprint to the nearest control then figure out the rest along the way" - this might be a recipe for not identifying the optimal route. Some course planners might want you to do exactly that and would set a close control as a trap.
"If it's a short hour event you can not afford too much thinking time I like to be moving in 10-20 seconds for an hour event." - planning is 50% of the work in Score-O. Another minute of planning might be worth many more points.
More insights anyone? Let's keep this discussion going.
"Points on the periphery of the map can be connected"
"Make use of elevation - don't lose what you've earned"
"In very hilly areas especially, look for ways to string controls of similar elevation together."
"Quickly scan the map for impassable obstacles"
"Follow a main trail/road during some of the hours of darkness, picking up controls on both sides."
"Avoid acute angles in your course"
"...option of doing it one way round or the other."
"I have a timer set to beep...at half time"
And I love the idea behind"finger test"
A couple of comments that I am not sure that I agree with (disclaimer - this is just my opinion):
"Sprint to the nearest control then figure out the rest along the way" - this might be a recipe for not identifying the optimal route. Some course planners might want you to do exactly that and would set a close control as a trap.
"If it's a short hour event you can not afford too much thinking time I like to be moving in 10-20 seconds for an hour event." - planning is 50% of the work in Score-O. Another minute of planning might be worth many more points.
More insights anyone? Let's keep this discussion going.
Aug 17, 2020 7:02 AM
#
Question. Is it OK (or even standard practice) to design a course with two sets of controls spread over same area, set A and B. And say teams must take controls from either set first and after starting to collect controls from the second set they can't get points from the first set any longer. This would allow planning longer races on smaller area/map and use best area more and nasty parts less, less need for permissions from land owners for smaller area etc. Deciding when and where to change the set would change planning a bit, so would competitors dislike it and see this as no-no?
Aug 17, 2020 7:13 AM
#
@Jagge
That happens quite often in the UK on smaller areas, generally for low-key summer events.
That happens quite often in the UK on smaller areas, generally for low-key summer events.
Aug 17, 2020 7:58 AM
#
Can you point links to some maps to see how maps and control numbers look like (A51, B79 maybe) ? Thinknig of suggesting trying that here.
Aug 17, 2020 8:18 AM
#
@Jagge
Quite often it is odd numbers first, then even numbers (or a variation on that). Once you move from odds to evens you can’t go back to odds. Works well on small areas.
Quite often it is odd numbers first, then even numbers (or a variation on that). Once you move from odds to evens you can’t go back to odds. Works well on small areas.
Aug 17, 2020 9:13 AM
#
Odds then Evens or something like that: interesting.
Is there a timing program that accommodates two possible random choices within one competition?
If so I'm also in to give it a go.
Is there a timing program that accommodates two possible random choices within one competition?
If so I'm also in to give it a go.
Aug 17, 2020 9:35 AM
#
South Ribble OC in the UK have a New Year's Day score event every year with odds and evens controls eg https://www.sroc.routegadget.co.uk/rg2/#335&co.... The actual printed map has purple control circles for the odd controls and blue control circles for the even controls making it a bit easier to distinguish.
For NYD 2017 they went a bit mad and had 7 sets of controls based on the 12 days of Christmas or something like that, with the same idea that you could only start each group once and not go back to a group. Most people seemed to agree that was a bit too complicated.
They must have software that supports this format.
For NYD 2017 they went a bit mad and had 7 sets of controls based on the 12 days of Christmas or something like that, with the same idea that you could only start each group once and not go back to a group. Most people seemed to agree that was a bit too complicated.
They must have software that supports this format.
Aug 17, 2020 11:10 AM
#
Slow person's version of @Nixon scheme: keep finding controls until you've had enough of orienteering for today, it's just for fun, so don't worry if you're late*.
*Except, don't stay out after course closing time.
*Except, don't stay out after course closing time.
Aug 17, 2020 12:21 PM
#
Ifor:
We use Odds and Evens fairly often for small areas on our night events. No special software but it's a fairly easy manual check with typically 330-40 people. If someone has had issues and swapped back and forth I tend to work out what gives them the best score with just the one change. You do need to have a system that shows you second punches of the same control.
Map printing is a bit more varied for the format. Different colour circles helps. Back to back sheets are also sometimes used.
When running this sort of event the choice of the point to change over becomes very important.
Map printing is a bit more varied for the format. Different colour circles helps. Back to back sheets are also sometimes used.
When running this sort of event the choice of the point to change over becomes very important.
Aug 17, 2020 12:25 PM
#
the individual points of each control is important to determine a good route.
Years of score / rogaine results in my area suggest that even with a 2:1 or 3:1 point spread between high and low, the overwhelming predictor of finish place is control count.
Convinced of that, I just draw a line through them all that has a ladder shaped out and back loop for the end (easy to drop and know return time) and go for it.
Years of score / rogaine results in my area suggest that even with a 2:1 or 3:1 point spread between high and low, the overwhelming predictor of finish place is control count.
Convinced of that, I just draw a line through them all that has a ladder shaped out and back loop for the end (easy to drop and know return time) and go for it.
Aug 17, 2020 12:50 PM
#
One variation for a small area is a "dog bones" course. Every control is paired with another, generally with a connection line between them. In order to get points for either control, you must visit both of them (in either order), with no other controls in between.
You can also just give points for every control collected, but give a bonus for each bone (pair) collected without intermediate controls. This variation forces those who want to maximize score to go for the bones, while slower, more recreational people still get an opportunity to get a high control count with much less distance.
You can also just give points for every control collected, but give a bonus for each bone (pair) collected without intermediate controls. This variation forces those who want to maximize score to go for the bones, while slower, more recreational people still get an opportunity to get a high control count with much less distance.
Aug 18, 2020 4:01 PM
#
So many Score-O events let the competitors study the course before the start. I would always prefer that maps not be turned over until the start whistle. Studying the map prior decreases the challenge and thus enjoyment. Thinking under time pressure is part of the essence of orienteering.
Aug 19, 2020 1:37 AM
#
Score orienteering is generally dull until the last ten minutes, when it gets far too exciting sometimes.
Oct 20, 2020 1:41 PM
#
Another data point....get as many as you can. Spread was 30-90 pts per flag.
Oct 21, 2020 9:31 PM
#
One idea which I haven't seen mentioned so far is to try to take major climb on paths, and then pick the controls mostly downhill.
My last score-O was a night race SE of London a couple of years ago, there I messed up by not realizing early enough that I could have taken all of the controls. :-(
My last score-O was a night race SE of London a couple of years ago, there I messed up by not realizing early enough that I could have taken all of the controls. :-(
This discussion thread is closed.