What are the disatvantages of a random draw?

I cannot see any reason not to do it random, it seems to be the fairest way, and can keeps things interesting and means no-one can possibly have a problem with the draw. The unpredictability of a random draw is exactly what we want!

I dont think there is any real need to keep our draw system in line with a world standard either, we should do what works best for us.

anythings possible in regards to how we do it.
its just much easier to do it randomly in the program.
You don't have to keep track of a tree, just a pool.

Pro's as i see it

• helps stop the state Vs state thing
  
• only ever have 1 bye in a tier.
  
• makes the competition more interesting from competitor POV (you have the dread of selection every round what could be more fun?? ;->)
  

I thought there was reasonable agreement that the first round draw for the big interstate events would be set up in such a way that bots would be drawn against bots from another state.
After all it’s a long way to go to just fight the same few bots you battle against in your state series.

As far as byes vs. 3 ways it’s not that I like bys but they are a lot better than 3 ways in my opinion.

The Kerro ranking doesn’t work on 3 ways fights either.

As for additional random draws through the event I don’t see the advantage.

hrmmmm
thinking about the draw mechanism

for selection of a fight
pick random bot
see what state its in
select another bot at random from the pool provided its not in the same state, if no other bot is found then another compeditor from the same state is selected.
if there are no more bots then that one gets the bye.

I don't think this is optimal though.
because you could have a situation where with one big and two small states you have the big state fighting itself and the two little ones fighting each other.

Hmmm i wonder.
i think for 20 bots, theres 20x20 possible combinations?
so theres 400 combinations.
if we generate all possible combinations, for each one we count the number of state Vs state.
then from the results with the lowest number of state Vs state we pick a draw at random.

thats some more heavy duty programming lol.
_________________
Mechanical engineers build weapons, civil engineers build targets

3 ways are out, They are majorly unfair on any bots that have to compete in a 3 way fight compared to a one-on-one, as well as complicating results. If people want to run 3 ways for grudge and exhibition matches fine, but the elimination tree must be consistently fair.

I dont mind the seeding of state vs state in the first round if the number of bots is roughly equal. I would say put the entrants into "state pools", then draw state vs state until the pools are empty, so the last drawn may have to fight a bot from their own state if the numbers dont match.

I see that randomising between rounds reduces the potential number of byes, and keeps a bit of anticipation in "Who will I draw next ?", but I dont like changing the way the competition is run between events without majority agreement.

Dont most elimination tournaments (in any sport, racing, football, etc ?) run a pre-determined tree ?
_________________
Great minds discuss ideas. Average minds discuss events. Small minds discuss people

at the last robowars there was alot of state vs state after the first round
the scoopy Vs IG and cobra Vs Plan-G spring to mind.
Basically this way if its at all possible to avoid a state vs state fight then its done that way.
_________________
Mechanical engineers build weapons, civil engineers build targets

As far schelduling the event, radamizing each of the rounds will only slow things down. The time needed to redraw all the fights, making sure that each that we will get interstate fights, as you ponted out as a pro, will take quite a bit of time. Then there is the factor that there won't be a set out fight time that people can perpare for so will need to give 30 minutes between the draw to the start of the next round. With the current setup there would be no time lost and more fighting.

As for historical referances, when I have been trying to dig through information the find the fight trees for old events I have found it almost impossible to re-build trees from NSW randomised events in a format where I am 100% sure I have it correct even if all the fights are listed. Where as I have built trees from the Vic events when I have been given only half the fights and still managed to get it correct.

It seems simplier all round to keep the standard tree structure and that was the way I was drawing up the time table.

Mmm, well, I see the rationale, but havent got time to think through the possible ramifications, so lets hear some opinions from the other competitors.

Mr Kero's ? You guys know a bit about this sort of thing with your study of the bot ranking process, Do you see any problems with a randomised round/tier draw ?
_________________
Great minds discuss ideas. Average minds discuss events. Small minds discuss people

Haha, beat you by a minute.

They do, but they don't seem to take into account local grouping.

I think its also done for schedualing reasons, they run the bouts continiously and a tree works well for that.

But robot wars dosent seem to work well in a continous format. Too much stress on us as builders and dull for the audience.

Hey we are robot people, we are exceptional to start with why do we gotsta be the same as everybody else lol. If this system is better, we can use it and others will move over to it.

what do you think of the batching idea as a whole?
the total duration of an event will increase somewhat is about the only downside that i see.
(but its probbly offset by not delaying matches with last minute repairs etc)

dad also had a good idea. marshaling
so at 2 fights you are given a marshaling call. you go get your bots and put them in the marshaling area get everything ready.
at 1 fight you do your final setup
at conclusion of the fight you call the next bots to marshaling, play the intro tapes etc for ~1 minute while you change the bots and the judging is done.
then you start the next round.
I think we can aim for a 4 minute cycle time, some rounds will go over but there will be tapouts and forefits etc.
_________________
Mechanical engineers build weapons, civil engineers build targets

I was thinking of changing the format and removing the 30 minute thing totally.

basically we conduct all fights in one batch.
everybody gets their bots ready, then all the fights are run in one hit.
one after the other.

then you stop for a full hour, no fights at all for an hour or so from the completion of the last group of fights.

then you run the next tier.


Re: Draw time, i estimate a draw will take about 3 seconds, from the time you press the "draw" button to the time its all finished (VB sucks at array processing)
the previous NSW rounds were pretty hap hazard except when we had a dedicated organisey type person.
All reccords will be kept digitally so there shouldnt be any issues with re-constructing things. Any public facing information will come from the same DB (or be converted from) we use on event day.

Note, we should make a backup of that DB too on event day before packing up, its kinda important lol.
_________________
Mechanical engineers build weapons, civil engineers build targets

quote:

---

Originally posted by Valen:
everybody gets their bots ready

---

As far as I now all other sports that use a elimination tree do it the same or very simular to the way it was done with RoboWars events in the past.

If you have random draws between events then there needs to be min of 30 min between the finish of the one round to the start of the next.

For the RoboWars events this is not possible as they are tight on the schedule already.
Remember there is a time factor and RoboWars held events do not have spare time.
As Brett does an absolutely fantastic job getting so many fight in the allocated time slot..

Personally I think a random draw for the first round.
Loser of fight moves to first lose tree.
Winner proceeds to next round of winners tree.
No filtering for states.
(Filtering not as much of an issue for me. I do understand reason for having it, and agree to some level.)

Reason for no filters is
When filtering states, the second round could still have a high % of state Vs State.

I think some test might be needed for this. I'll look in to it soon.
Have a lot of rubbish going on at the moment and need to be sort out first.
_________________
It's all about the destruction.

with 20 bots i came up with 19 hours with 1 hour between rounds.

http://www.robowars.org/forum/viewtopic.php?t=758&postdays=0&postorder=asc&start=24

if everybody has an hour to prepair then i dont see any difficulty with an "everybody be ready mandate"

you could presumablly abandon the whole "1 hour gap" and go for a more concurrent system with some clever schedualing (ie add the minimum 30 minutes to the requirements for battle order)/

it does seem like a fairer system than the concurrent though.

(theres also a maximum of 1 bye per round in the pool/random system)

edit
you can run the lightweights and demos in the gap periods.
I was hoping that the long gap between tiers would encourage more "for points" battles.
theres another thaught, run your ranking system (or some other simmilar type thing) then people can try to also be the winner "on points" or some such, I'm trying to keep people battling with a purpose/incentive every hour or so rather than just eliminating people from the comp, which always leads to the time crunch at the end
_________________
Mechanical engineers build weapons, civil engineers build targets

So you have 3-ways/bye right through to the end of the compitition? That ain't fair really. You need to remove the byes or three ways out of the tree a early as possible. With the radimised method of always suffeling competitors to the top of the tree and pushing the byes to the bottom you will always have the byes being filltered down into the last rounds. With 24 robots in a 32 robot style compitition tree if you move the byes at the bottom the winners side will end up with 4 spots and 3 robots at the end. Who will be the lucky bastard to be garenteed to make the top 3 and into the money? Admit it, thats how the other two compititors will feel.