A person can get "randomly" lucky and end up with a top score, but in general, the players who have the highest score in a given episode or app got there by determination and skill.

Ask them about what? If they built "random chance" into the app? Even if they did, no developer would ever admit to it.

Twitter answer from Rovio is "Squawk! A bird never tells" : (

The largest randomizer is perhaps the finger controlling the slingshot. Without the ability to measure angle or position with a repeatable and reliable standard, say X/Y coordinates on a pixel scale, every pull of the slingshot is but an approximation of any example given on any level. Thus, as merely rough copies, different users will achieve different results with what SEEMS to be a good replication of the shot.

I haven't played with the Box2d physics engine yet, but reading some of the documentation has been enlightening. Just about every variable you can think to apply in a 2d space has been accounted for with specific limits, which can be adjusted by the end programmer. Exceeding certain value ranges on certain program sprites can lead to a slight derangement of the program as it calculates destruction levels and point totals.

Levels that require, or seem to require very specific angles, locations and bird arcs, represent where the programmer has tightened up some of the allowable ranges for specific reasons. Some are strategic in nature, and some are programmatic, to accomplish what the game designer had in mind. And some are simply accidental.

AB is a case of random elements (i.e., YOU) driving the physics. What the game designer and programmer have done is to set up initial conditions from which all else follows.

Has anyone seen this?
https://www.facebook.com/photo.php?fbid=10150405827249928&set=pu.314467614927&type=1&theater

Not a whole lot there, though it looks confusing a first glance. Most of that you learn in a first-year high school physics class: velocity, acceleration and path of a parabolic projectile, work exerted on/by an object, etc.

Why would Rovio not acknowledge whether there was "random chance" programed into the game or not. What difference does it make to us/the players in them not letting us know?

I don't think any game programmer would admit to designed randomness (with the exception of obvious circumstances). Programmers pride themselves on logic and consistency.

That said, I don't believe there is any randomness to Angry Birds (again, with obvious exceptions such as the designed randomness of awards in AB-Facebook).

In terms of 0-bird levels, or "Rube Goldberg" levels, I think natural variation occurs in all levels of all games. Structures are slightly suspended above the ground, and as they fall that 1 or 2 pixels, blocks must bounce and settle. Variation and randomness are not the same thing: Variation is expected; Randomness would mean a designed "random generator" counter, the result of which would determine what is going to happen.

@TruckDriver, the drop observed in most levels is not a glitch, but simply due, imho, to the fact that the physics engine uses the MKS system ( Meters, Kilograms, Seconds ). AB uses pixels. The game code must convert those numbers so that they're useable by the engine using a conversion factor. Like most conversions, the result is never a round figure.

All the techno talk usually makes my grey matter hurt but I actually think I understand what you said TruckDriver. Basically the randomenss is a result of the old "lost in translation" and putting it that way, it makes sense. Therefore I believe what slim said re Rovio not deliberately putting the randomness in but that it just happens.

What causes the variation? If people agree there is some variation in the starting conditions (a finite, non-random selection), don’t you think the choice of the starting conditions is random?
I have written many programs to solve physics problems (but no game programs), and have a different perspective:
Perhaps the term random is too technical. Speaking as a game player, not a scientist (which I am), I think having a number of different starting points to a level is what most people would classify as “random”, even if there are only 2 or 3. And what determines which of those 2 or 3 you get? So speaking non-technically, random = variation.
@loox – I don’t think your basic proof has really been answered. I agree that differences in the starting conditions (pre-settling) would be easier to implement than adding a random variable generator to Box2D. There could be some small coefficient of variation to position / orientation, rather than variation added to physical characteristics. It would, however, help (me at least) understand what is going on in AB if you can remember specific levels where spontaneous collapse “randomly” occurs. I remember seeing these, but can’t remember which.
@TruckDriver – Agree that the largest random variable is the user. Variation in the outcome of the initial settling is what most users would characterize as random. What causes the variation? Why would a program sprite appear at a different time? If a program sprite affects the outcome of a level, that would mean that timing is critical, and I don’t think most AB players consider that, except in levels where there is obvious motion (e.g. things swinging). Does Box2D have program sprites (I didn’t see them, but may not have looked carefully enough)? What do you mean by “derangement of the program”? I hope we are not at the level of machine error here, and not initializing variables properly is a programming error. I agree that no one would randomize starting variables, but they could add a small amount of randomization (e.g. variable*(1 + (0.01x random number from 1 to 10))).
@Burbman – Your supposition that there might be a number of release points satisfies most game-players definition of random.
@Slim – Agree that variation and randomness are not the same, but variation also needs to be programmed in. What determines the variation in the starting position?
@Sal9 – The conversion might not be a round number, but it should always be the same (and I expect Box2D uses decimal figures).
BTW Rovio still hasn’t answered my e-mail (they answered my tweet).

@mvnla2 - all physical data of a level are saved in a finite number of decimals. Let's say one specific block is located at 3.562 meters from the left bound. In physical means, that is any value between 3.5615 and 3.5625 meters, so in the first try that could be 3.5618 meters, next time maybe 3.5624 meters (who knows?). These tiny differences can have big consequenses (the butterfly effect?).

And BTW, the calculations in that pic are VERY wrong (with an accelleration of 9.8 m/s^2 the bird would fall upward for example)

New physics in SaT? Well, not really new physics, but maybe some new boundary / initial conditions that could make some levels even more difficult. SaT series 1 has waves; 2 has a waterfall, a river, and maybe wind; I bet 3 will have rain (maybe even thunder and lightening). Did Rio have waves? Haven't checked, and haven't noticed that the waves affect results so far. However, look at SaT 30 and that swing. Under some circumstances it continues swinging long after the pigs are dead, and even starts making larger motions. Only physics-based explanation I can think of is that there is some wind, possibly influenced by the waterfall. Introduction of weather should make for some very interesting, not to mention frustrating, gameplay.

@loox Thanks for the response. I am working on a reply (which I hope you like).

Much has been said here with regards to random chance and the AB physics engine, but, with respects to all, I ain't buying today. Knowing what I know, I can't.

User perception of a program at runtime is a metric of extremely limited value, since most of what occurs in AB, or for that matter, any software, is running under the hood, out of sight of any user. The mere appearance of randomness cannot be inferred as true randomness simply because we witness a behavior that we did not expect to see, or lack the ability to measure and quantify.

Every element within the program is assigned specific values, for specific physical properties, in a simulated two dimensional space. Every interaction within that space occurs under defined constraints, at an exact rate of occurance, which far surpasses our human ability to even see, much less correctly process.

Theoretically, similar elements on one of the game levels, for example a narrow board, should possess exactly the same set of values attached to it, as every other appearance of that same element. Initially, when each new level is designed, I have no doubt that each iteration of the same element are indeed possessing equal values. That brings us to the testing/debugging phase of the development process.

Part of the rub is that each individual level is designed to be completed with its own specific method, with a defined range of end score point values to stars earned for that level. However, if testing reveals a problem with the levels as designed, then those values of the problematic element(s) are adjusted until they obtain the objective desired by the game designer.

Then the levels are meshed with the rest of the game code which is again retested, to look for, then fix, any other problems that might have developed. Also tested is overall crashworthiness across all the end user platforms for which the game is designed/scheduled to be released and run on.

Another part of the rub is that this process takes place over a span of weeks and/or months, and now, certain elements, despite visual similarities, might now have unequal values. Still another is that no testing regime is going to uncover every interactive flaw that can possibly occur.

Lastly, the proofreading of program code is a difficult, tedious process, wherein mistakes or errors can be overlooked, especially if no overt problems are detected.

This leaves us with a set of elements which started out equal, but through the design process, may in fact, no longer be equal by the time the program is released. This may be through design, compromise, or omission, but present nevertheless.

Could this be the case with AB? Probably. In fact, almost certainly.

Is it random chance? Unlikely. At least not within any definition of the concept that I've ever come across. It isn't even much of a concept, except as a sort of catch all to explain phenomena that we don't understand, or can't precisely quantify. The concept of random chance is insufficient to explain random chance. The pot cannot hold itself.

Finally! Someone gets what I've been saying!!!

To make it clear one last time (because I think @truckdriver and probably others are making a subtle misinterpretation of my point): I'm not saying AB has randomness. Rather I am merely saying that because on the levels with spontaneous destruction, like AB: Seasons, Wreck the Halls 10-5, it is impossible to predict my score before throwing any birds (should be zero, but instead, and as @mvnla2 noted, its value changes). This is because the initial conditions differ every time.

If you read my post carefully, I say that this makes AB *seem* as if there is a random component, when actually it is more likely a parameter (or variable) that we haven't yet perceived.

Or it could be an actual random number generator, but (and I'm no programmer) I don't consider RNG's as truly random anyway since they have to be coded. Thus, they have to be based in some sort of algorithm with a changing variable, like time... But that's a whole different topic. The point is that if you can't predict the score of a level before throwing a bird, then something is going on that may as well be treated as random, because even though it's not truly "random", to the extent that you can't predict the score on WTH level 5 means that for our intents and purposes it can be argued that random chance is a factor (to name but one of many levels)... The rest is semantics. If this works in my favor, I think of it as good luck... Even though I don't believe in luck, per se. Call it anything you want, but this behavior is present in AB.

@truckdriver, much has been said about random chance because that's the topic of this thread. In any case, I think we are actually both saying the same thing... You turned to the code to support the notion of a variable we don't perceive. I used the initial conditions of certain levels as proof. But we both seem to agree that something in the code is causing slight variations... Something that, as of right now, we don't understand. Whether its a result of the physics engine or a variable introduced by the programmers is irrelevant to the topic of this post, which is "does your score depend on physics or random chance?"

I think that has been answered: it's physics, with a touch of what might seem like random chance, but is really a parameter, variable, coding or physics engine glitch that may as well be random because we can't control it or even pin down exactly what it is. Whether or not "random" is the technically correct word tor this behavior is irrelevant.

Either way the end result is the same: even if one could duplicate the same exact shot, the results may differ because the structure may positioned slightly different (or even a lot different) when one restarts the level.

OK, I started to say I may backtrack a little, but half a sentence later (deleted), backtracking doesn't make sense. We all agree that inability to precisely duplicate shots is not a random (or even variable factor in the sense we are discussing). When spontaneous destruction occurs, what starts the program in motion may not be random, but is sometimes variable. The programmed variation may not be random, but it is real.
As to the absence of data, the statistics I provided above, and which can be added to (an exercise for the student) are real data. Data can be based on quantified observations rather than knowledge and study of computer codes.

@Truck Driver: Like you, I am not looking for a consensus of opinion, just data. Considering a variety of opinions is useful to me in trying to distinguish between opinion and data.
At the most fundamental level, all I am saying is that for some levels, the starting conditions vary from start to start. Since I personally only accumulated statistics on one level, Wreck the Halls 1-5, all I can claim is that it happens on that level (loox claims it happens on several, and I believe him/her). Why do you consider the data I collected anecdotal? Granted it’s not publishable in a technical journal, but it is real data non-the-less.
Answers to your questions:
1. How would you provide a control, and under what conditions? The idea of a control in experiments applies primarily to experiments on nature (where there are many unknown/unknowable variables). From Wickipedia: “The important thing is to try and control variables and attributes in the data so that the conclusions drawn are valid.”
2. How verifiable is it? If you try to collect the same statistics on Wreck the Halls 1-5, I would expect you to get the same result (within experimental error).
3. How repeatable is it? The statistical distribution of results should be the same within experimental error. Granted, I haven’t calculated the experimental error. I could, but please don't ask -- this isn't a technical journal after all.
4. What would be an optimum sample size? Well, the larger the sample size, the more accurate the results, and the closer you would come to proving whether the variation was random or just some finite set of variations. My sample size was 40+, which I think is more than adequate for proving that there really is a variation in starting conditions on WtH 1-5.
5. Would it be OS and/or device specific? Most likely, because Box2D says the results of their program are device specific. So we could modify my conclusion to apply only to an IPad, which is what I use. However, I would postulate that the existence of variation on WtH 1-5 would still be obvious on other devices.
6. What would be the optimum number of levels, and from which version of the game? As far as I know, no one is saying that this variation in starting conditions occurs on every level or every version. All we are saying is that there are some levels where this occurs. I have personally only offered proof for 1.
7. What would be the acceptable range of player ability? If you are looking at starting conditions, with no birds shot (which is what loox and I are talking about), player ability is irrelevant.

ABOS 2-3 Sometimes the closest large asteroid is stationary, sometimes it rolls clockwise and sometimes counter-clockwise.