Forum

Search posts

 

Ilia Poliakov -

Siteswap difficulty program.

I thought, which siteswap is harder and why. And understand, that it is possible to create algorithm counting difficulty of siteswap. I think up one easy algorithm, and then try to find something in internet.
I found this: http://www.sporttaco.com/rec.juggling/siteswap_difficult_algorithm_3961.html
And this: http://www.giocoleria.org/manuali/Ben%20Guide%20to%20Juggling%20Patterns.pdf
First thing is not complete, but Ben's Beever system, I think, very good.

So, algorythm is good, but I want to find program, which would count siteswap difficulty. I want to have got a list with tricks sorted by difficulty. Who know about it? Or maybe you can do it for me? Just one formula for Excel.
Also, in Ben's system it is possible to give difficulty for every trick, like 534 Mills Mess with Claw.
Maybe we can use this for our site and range all kind of records automatically?

peterbone - - Parent

I think that rating all patterns automatically would be very ambitious. Ben's formula is good, but not perfect.

The way I would approach it would be to use more of a simulation method rather than a purely mathematical approach (I'm an engineer). Simulate a juggler and the props in the pattern as accurately as possible using a physics engine. You could then measure the amount of energy used, the proximity of the props to each other, the chance of collisions and so on. Adding small throw errors would give an indication of how much energy / accuracy is needed to make corrections. Combining all this information would I think be the only way of accurately rating difficulty for arbitrary patterns with a range of props.

Ilia Poliakov - - Parent

Do you think, somebody will do it in this century?
I don't find a perfect formula. I think, we can correct Ben's formula, include some for collisions.

emilyw - - Parent

Can Marvin not do it?

Marvin - - Parent

In toss juggling all siteswap patterns merely involve throwing the props to the correct height at the correct time which is a trivial task providing you can physically throw to the maximum height required. Therefore all patterns that have a maximum throw height within your physical limits have a difficulty of 0, all patterns involving a throw height outside of your physical limits have a difficulty level of 1.

Orinoco - - Parent

Marvin, get back to untangling all those cables.

emilyw - - Parent

Oh Marvin, you're always reducing problems to ones and zeros!

Ilia Poliakov - - Parent

I make it for OpenOffice, but if it is posibble to make shorter, it will be great: (A2 - siteswap)
=2+SQRT((IF(LEN(A2)>0;(MID(A2;1;1)-2)^2;0)+IF(LEN(A2)>1;(MID(A2;2;1)-2)^2;0)+IF(LEN(A2)>2;(MID(A2;3;1)-2)^2;0)+IF(LEN(A2)>3;(MID(A2;4;1)-2)^2;0)+IF(LEN(A2)>4;(MID(A2;5;1)-2)^2;0))/LEN(A2))

Orinoco - - Parent

For Excel users: replace all ';' with ',' to make the formula work.

eg:

=2+SQRT((IF(LEN(A2)>0,(MID(A2,1,1)-2)^2,0)+IF(LEN(A2)>1,(MID(A2,2,1)-2)^2,0)+IF(LEN(A2)>2,(MID(A2,3,1)-2)^2,0)+IF(LEN(A2)>3,(MID(A2,4,1)-2)^2,0)+IF(LEN(A2)>4,(MID(A2,5,1)-2)^2,0))/LEN(A2))

yDgunz - - Parent

I think there are 2 ways to assess the difficulty of patterns. You can compile people's opinions and rank the patterns accordingly, or you can try to come up with some algorithm to generate a numeric difficulty score (basically what you're suggesting). At the end of the day I'm less interested that 7531 has X difficulty score, and more interested that 7531 is harder than 534 but easier than 7441. The value in that would be a progression ladder for folks exploring new siteswaps. You could combine these ideas by testing the algorithm against people's opinions.

I like Peter's idea of apporaching this from a physical simulation point of view, as opposed to just assigning weights to pattern components and applying some formula to them. Toss juggling boils down to making accurate throws in both time and space without collisions. Collisions happen when you introduce error into a throw. It would be interesting to assess the impact that error has on whether or not a pattern will collide. For example, you could observe that errors in 5-ball cascade have X times the impact as errors in 3-ball cascade (I feel like someone has already done this...) You could introduce different throw/catch types/locations by assessing their impact on throw accuracy. I think this would be more experimental and is really going to vary by person - for exmaple, I've spent a lot of time on fork throws (because I enjoy contact juggling) and have pretty much ignored over the head. So of course that's going to impact the difficulty of the pattern for me.

I think this is a really interesting topic. Are there other resources out there besides the links that Ilia posted? Ben Beever's book is pretty awesome, very comprehensive but not so complicated that you need a math degree to follow along (there are a few juggling related papers I've tried to read that fall into that category).

yDgunz - - Parent

And I just tested my opinion that 534 < 7531 < 7441 against the Excel formula and perhaps I was mistaken? Apparently 534 is 4.2 difficulty, 7441 is 4.9 and 7531 is 5.0. Granted they are pretty close, but I guess I should keep working on 7441 then!

Brook Roberts - - Parent

I certainly agree with the formula there - 7441 is definitely easier than 7531 - it's much easier to do a very fast low 441 than 531.

Ilia Poliakov - - Parent

I think,you will agree, that 333<441<531. So, 7333<7441-7531.
Simple logic)

^Tom_ - - Parent

7441 is much much easier than 7333.

Of those three patterns, 7441 is the easiest, then 7531, then 7333 in my opinion.

Sjors Stuurman - - Parent

I think I'll agree with Tom that a 7441 is easier to fit into your pattern then a 7333, I've recently picked up 4b siteswaps again and I feel that it is way easier to fit a quick 441 in then to do 333, as to 7531, that's really more difficult.

Ilia Poliakov - - Parent

I will try in monday.

Ilia Poliakov - - Parent

Yes, I always did 73334, and never 7333.
7333 is harder then 7441.

Adrian G - - Parent

I like the idea of compiling what people think (especially as there are so many differing view on tricks that formulas will never be very accurate). I seem to remember an idea being mentioned at one point where people don't have to rank a whole list of patterns but just getting a lot of people to answer 'is X or Y harder?' and combining all of those into a somewhat decent ranking of tricks.

I think also a lot of people mix up comparing running one pattern to doing one round of another out of a base pattern. For example I find doing one round of 7531 easier than one round of 7441 but find running 7441 easier than running 7531.

However, all of that is still vulnerable to certain tricks being practised more, unfortunately...

There are a lot of suggestions in the rec.juggling archive, if those count as 'other resources'. I don't think it's something that many people have really put a large amount of time into to write up about in a paper though, probably because it's so subjective. I don't think the Mathematics of Juggling looks at difficulty but that's the other obvious resource to look at.

Ilia Poliakov - - Parent

One important thing - ask people about tricks, in that they have got records.
If I have got record in 4b mills mess and 4b 534 mills mess, you can ask me about comparing this trick.

Mike Moore - - Parent

In comparing the difficulties, are you talking about difficulty to flash the pattern, qualify it, or run many catches of it? I've heard from many people that running 5c backcrosses is easier than running 5b backcrosses, but I bet a flash is easier with balls because recovering into a 5b cascade is so much easier than recovering into a 5c cascade.

Ilia Poliakov - - Parent

maybe we can ask 3 times?)
I don't know, what is the most important - flash or qualify or many cathes.
If I choose many cathes, for some tricks it is unreal.
If qualify - for one trick it is 6 cathes, for another - 12 cathes. I think it is not good system to compare.
P.S.Mill's mess qualify - is 6 or 12 cathes?

Little Paul - - Parent

Mill's mess qualify - is 6 or 12 catches?

Clearly 12 - as it's a 6 beat pattern to get you back to the staring position, and a qualify is "2 cycles" of the pattern.

It all gets a bit mad when you start trying to apply mills mess arm movements to 4 beat siteswaps though.

Ilia Poliakov - - Parent

Yes, 5 ball mills mess and 6 ball mills mess - qualify is 12 catches?

I don't think that flash gives any information about trick. Flash 4 clubs flats is easy for me, but I can't juggle them more.
And I don't think that 100 cathes is not good because it means, that we can ask only people who can make 100 cathes. For some tricks it is too much.
I think, that most of jugglers who like to beat records use WJF qualify system (what else they use, I don't know. They give more points for more then qualify cathes? They give pointes for flash?)
And I will choose Qualify. But always we need write "Qualify: 5b cascade(10 cathes) or 4b mills mess(12 catches)"

What do you think?

^Tom_ - - Parent

I believe that the WJF awards no points for flashes. I thought that went some way to explaining the song that Luke Burrage set his routine to.

loganstafman - - Parent

I've been thinking about this as well, as I've been hacking up JugglingLab recently. Having juggling lab allows you to get information not present in the actual numbers of a siteswap pattern. For example, how close balls come to each other in the air could help determine which siteswaps have issues with in-air collisions. I know there's more info that can potentially be gleamed from this method.

Rob van Heijst - - Parent

I'll just leave this here:
http://www.simonacampo.nl/Profielwerkstuk.docx
http://www.simonacampo.nl/showcase/b21/Moeilijkheidsgraad-calculator.html
I've asked Simon he can help a bit more maybe. He also says that his algorythm isn't perfect and that their can't be a perfect algorhytm.

Ilia Poliakov - - Parent

In my browser it does not work.
I dont understand Niederland language, but I understand idea.
So there are some good ideas, but we have no anybody, ho realize a good system.
I think, we are very close. And we have good very good instrument to do it.(jugglingedge and you,guys)

For me it is not important - to make a program, or ask people. The mix of this ways should work.
It is very important, that we no need a numeric score, but we want to see a compare list.
I think, it is stupid question about difficulty from 1 to 10. (juggling-records.com)

We can ask people to compare two tricks.Than other two tricks, and other. Some details are important - like to do this trick only ones, or make 100 cathes, or...
What do you think? Very easy, funny and useful.

Ilia Poliakov - - Parent

or maybe we can use all records on site?
Everybody has got a great collection for comparing. If I have got 314 cathes in 741 and 227 cathes in 714, then we have got 1 vote, that this trick harder in 314/227 times. Other guy has got another situation - 23 cathes of 741 and 37 of 714. So, now 714 easier. Calculate 314/227 * 23/37.

peterbone - - Parent

The problem is that some tricks are much more popular than others because they look nicer or for some other reason, so people train them a lot more.

Little Paul - - Parent

Yeah. The problem with using records as data are that you "get better at what you practice"

To take a simple example, most people would agree that 4 balls is easier than 5. However, then you get someone like me who comes along and never bothers to learn 4 balls, never practices it, I skipped it and worked on 5 because I was more interested in being able to juggle 5 balls. I don't record my records, but if I did, and you took my personal bests with 4 and 5 as data, you'd look at it and see that I get more catches with 5 than I do with 4, so you might conclude that 4 balls was a harder pattern than 5 balls.

It's not, I've just logged a couple of orders of magnitude more hours working on 5 balls than I have on 4.

Peoples records say as much about what they're interested in working on as they do how hard those tricks are, and without the missing "how many hours have you worked on this trick" data (which hardly anyone ever records accurately) you can't really use them to judge difficulty in any meaningful way.

Ilia Poliakov - - Parent

Yes, its true. But I want to see, what happend if we will make this program.

Ilia Poliakov - - Parent

Another idea - if we will take world records and will compare them.

Ilia Poliakov - - Parent

Last change - now it works for patterns with symbols (like b4a33333) and LENGHT=12
B2 - siteswap
=2+SQRT((IF(LEN(B2)>0;(HEX2DEC(MID(B2;1;1))-2)^2;0)+IF(LEN(B2)>1;(HEX2DEC(MID(B2;2;1))-2)^2;0)+IF(LEN(B2)>2;(HEX2DEC(MID(B2;3;1))-2)^2;0)+IF(LEN(B2)>3;(HEX2DEC(MID(B2;4;1))-2)^2;0)+IF(LEN(B2)>4;(HEX2DEC(MID(B2;5;1))-2)^2;0)+IF(LEN(B2)>5;(HEX2DEC(MID(B2;6;1))-2)^2;0)+IF(LEN(B2)>6;(HEX2DEC(MID(B2;7;1))-2)^2;0)+IF(LEN(B2)>7;(HEX2DEC(MID(B2;8;1))-2)^2;0)+IF(LEN(B2)>8;(HEX2DEC(MID(B2;9;1))-2)^2;0)+IF(LEN(B2)>9;(HEX2DEC(MID(B2;10;1))-2)^2;0)+IF(LEN(B2)>10;(HEX2DEC(MID(B2;11;1))-2)^2;0)+IF(LEN(B2)>11;(HEX2DEC(MID(B2;12;1))-2)^2;0))/LEN(B2))

Orinoco - - Parent

Or this form supports siteswaps of any length with throws up to height z.




Siteswaps 0, 1 & 2 obviously have anomalous difficulty ratings using this formula.

Any formula/algorithm/statistical analysis will only ever end up with an approximation. I've never seen the point of analysing complexity personally. What does it achieve? What can you do with a list of incrementally difficult tricks? Is learning progressively difficult tricks optimal, interesting or otherwise beneficial?

I choose what tricks I want to learn based on what I find enjoyable to do & what looks good. I never tell people I'm teaching how difficult I think a trick is either as that just puts up a barrier to learning.

varkor - - Parent

I like how it thinks (6x, 4)* is harder than the 14 ball fountain! I feel rather proud of my efforts now. It does make you wonder why Alex hasn't managed to flash 14 yet, though...

Orinoco - - Parent

The button text is accurate!

I should've mentioned that Ben's formula covers Vanilla Siteswaps only & that little widget doesn't do any validation.

Ilia Poliakov - - Parent

Do it yourself)

Ilia Poliakov - - Parent

It gives me a list that can help me to teach people and it is good for me if i think - which siteswap I can learn? The best way for me - is learn tricks, that hard for you, but possible to do them in 5-10 minutes.
If I will try to do b97531 now, it will be not good idea. And I know, that a lot of people don't understand that trick they want to do is very hard for them. But they try and try and try again.
At the same time I can learn 5-10 easier tricks and start to learn the same trick.

Little Paul - - Parent

These days I'm more interested in learning individual tricks which are so far outside my skill set that I'm pretty certain I'll get nowhere quickly with (but then I'm not a toss juggler)

My current obsession is mouthstick+bottle, which I started working on in September and in the last week I've managed to catch a throw onto the mouthstick twice (in two practice sessions) and I'm super stoked about that as progress.

Ilia Poliakov - - Parent

What do you mean? In general.

Little Paul - - Parent

I don't really juggle balls/clubs/rings any more, I've been juggling for about 22 years now and I'm just not interested in them any more - so rather than working my way through an endless list of siteswaps I'm working on other props.

You said that you like to work on tricks which are just beyond your reach, but which you can learn within an hour. That doesn't satisfy me any more, and I prefer working on tricks which are *way* beyond my abilities. Tricks I don't expect to be able to learn in under 3 months.

Everyone is different, and I just find tricks like that much more satisfying to learn!

I spent probably 3 months working on The Golf Club Trick on and off until I got it solid enough that I'd happily attempt it on a renegade stage, but my current goal is a sequence which appears to be far harder to learn, but which looks much less impossible.

For example, the sequence I'm working towards at the moment works like this:

1 - hold a stick in your mouth (I'm using a wooden spoon)
2 - throw an empty wine bottle so it spins, and lands balanced on the stick
3 - balance the empty wine bottle on the stick
4 - using a flick of the head, throw the bottle up into the air, and catch it back on the stick

In September I started working on step 1, and step 3 (the easiest parts of the sequence) and within a month I had the balance pretty solid. So I started working on step 2.

This week, I've caught the throw twice, in my two most recent practice sessions. I'm not kidding myself that they're anything other than fluke - but I'm happy with the progress.

At my current practice rate, I think I'll have the whole lot by this time next year, probably not "performance ready" but at least "more often than not".

I guess I just get a kick out of putting a horrendous amount of work into learning a trick which fills at most 30 seconds of stage time.

The Void - - Parent

More knees.

Little Paul - - Parent

I do say that to myself repeatedly :)

Orinoco - - Parent

More wine!

barnesy - - Parent

best of both worlds perhaps: https://groups.google.com/d/msg/rec.juggling/7cyh6bJ_lvc/X4RuHBJ8_BUJ

Ilia Poliakov - - Parent

You practice only with bottle?

Little Paul - - Parent

I'm not sure what you mean, I thought I was quite clear about what I'm working on at the moment.

Ilia Poliakov - - Parent

When I train a balance, I take different stuff. Start from long and big, and then shorter and smaller.

Little Paul - - Parent

Ahh right. I understand you. When I was learning chin/nose balances I did the same thing, and worked down to around 8" (never did quite make it to a well controlled teaspoon balance)

Mouthstick balances are a little different to most balances. You don't control forward/back by moving your head forwards and backwards, you control that plane of motion by tipping your head forwards and backwards. With a wine bottle, you have 2 points of contact with the mouthstick, so you don't have to worry about that plane at all, only the side to side movement.

So to replicate that with a longer object you need something long and wide. I tried with a large cardboard tube (which happened to contain a promotional poster for "cards as weapons" I've have for almost 15 years but never got around to framing) but that hit another issue.

I generally balance by looking at the top of an object. It's useful if you can keep the mouthstick horizontal, but that limits how far "up" I can comfortably look. Which means that I was finding objects longer than about 14" more difficult than I had expected. Wine bottles are about 12" long, and 2" doesn't make much of a difference, so I gave up on long objects and went with the bottle.

In terms of "feel" it's a very different balance to the chin/nose or even the golfclub setup (which it shares some similarity with) which is why it took me so long to feel comfortable with it.

That said, I've been practising "over" the trick, by also working on the much harder football-on-a-mouthstick balance. I don't have a channel in my mouthstick (which would make it slightly easier, but cost me money) so I'm not really making a huge amount of progress there, but it's an interesting diversion.

Little Paul - - Parent

btw - the slow reply was because I've spent most of the last week or so reading the edge on my phone, and that doesn't really lend itself to long replies like I thought this needed.

Orinoco - - Parent

I generally balance by looking at the top of an object. It's useful if you can keep the mouthstick horizontal, but that limits how far "up" I can comfortably look.

Adjustable angle mouthstick?

Little Paul - - Parent

You've now got me wondering about building a mouthstick equivalent of
https://youtube.com/v/grflWD1zg-Y?start=18

Ilia Poliakov - - Parent

Thanks. You say so interesting things about this balance, that i want to try it)

Orinoco - - Parent

My ability to learn something is directly proportional to how excited I am about it. I am far more likely to get excited about something that is a long way beyond my current ability than something I know I could do with a little bit of practice.

In the video club juggling & passing Haggis recommended started to learn the 5 club cascade as soon as possible, "because it takes so long to learn". Given all the kids who seem to learn it in 5 minutes these days it should probably apply to 6 or 7 clubs but I still think this is good advice. Some tricks are so hard that they can take a sizeable portion of your life to learn, so the sooner you start the longer you will have to enjoy them.

The Void - - Parent

For a given value of "enjoy".

Ilia Poliakov - - Parent

I wanted to say, that it is stupid if you just try 5 club cascade without any exercises, like 55500,55550,552,50505. Also it will be good to be comfortable with 4 clubs and do fountain in single,double,triple. And 53 is good...
So, there are a lot of tricks that can help you. But sometimes people don't understand it.

Orinoco - - Parent

Stupid is a harsh way of putting it. There are many different approaches to learning. I never did any intermediate exercises for 5, 6 or 7 balls or 5 clubs. I've flashed 8 balls but I can't reliably do 4 balls in either hand. I find that practicing so called exercise patterns make me better at those exercise patterns but they don't really help me with the end result. I think this is because I tune my juggling to whatever feels/looks best to me for any given trick. This usually means that I will juggle a 5 object pattern with a hole to a different height/speed/rhythm to the way I would juggle the full 5 object pattern. For example I do 4 club 5551 on triples, but the 5 club cascade on doubles.

Ilia Poliakov - - Parent

I always say to my students - if you want juggle 5 balls, take 5 balls. Yes, you need start feeling 8 balls in the air. And you need adapt for them in your hands.

 

Subscribe to this forum via RSS
1 article per branch
1 article per post

Green Eggs reports