Things you often Wonder

[snowychap]

9 minutes ago, blandy said:

Sure - If you make one assumption, it does. If you make a different one, then it doesn't. One (your, valid)  assumption is that the question validity period is from the moment that any track is first played until 36 tracks have been played, another is (my, also valid) assumption that "if the current track being played is from the album, what is the prob. that of the other possible 35 tracks to come up (in the next 35 songs I play, with no repats), the three immediately next will all be from. the one album.

Which is back to where we started, this only works with your calculations if no tracks have already been played. And the numbers differ until 36 tracks have been played and then the probability is either 1 or 0 or unkowable (if we don't have access to the data concerning the 35 tracks before the one currently playing)

With a rolling 36 track cycle of no repeats within that cycle, if track N is playing, track N+1 must necessarily be the same as track N-35, surely?

Edited June 27, 2018 by snowychap

[blandy]

49 minutes ago, snowychap said:

Which is back to where we started, this only works with your calculations if no tracks have already been played. And the numbers differ until 36 tracks have been played and then the probability is either 1 or 0 or unkowable (if we don't have access to the data concerning the 35 tracks before the one currently playing)

With a rolling 36 track cycle of no repeats within that cycle, if track N is playing, track N+1 must necessarily be the same as track N-35, surely?

Ah, I see your point. The "no playbacks" caveat within a pseudorandom algorithm is tricky, if not impossible. And for a genuine random function, then you can't assume "no repeats".

 

[bickster]

Spotify definitely doesn't play all the tracks within a playlist before it repeats, even on a 30 track playlist like say Discover Weekly, on shuffle it will play a couple of tracks more than once before it plays some of the tracks it hasn't played.

There is definitely a number of tracks it will play before a repeat though and this is the number of tracks it downloads in advance for the next sequence of tunes, I've not managed to work out how many tunes this is yet though as I've not been stuck in a traffic jam in a long tunnel on enough occasions to get a full sense of it, what it does then is revert to the tunes it already has in memory but does not repeat them in the same order as it previously did.

Given that this is the way it operates, it could never be absolutely random, then again nor does it claim to be. Completely random would be awful I think. I do think it tries to group tunes it considers to be slightly similar together and then moves it along to something else after 2 or three tunes, I definitely notice it groups similar styles together a few songs at a time

[blandy]

2 minutes ago, bickster said:

Spotify definitely doesn't play all the tracks within a playlist before it repeats, even on a 30 track playlist like say Discover Weekly, on shuffle it will play a couple of tracks more than once before it plays some of the tracks it hasn't played.

I don't think any of the ones I've used don't repeat, either (e.g. iTunes, apple music whatever it's called). We've been using the no repeats to help with the statistical thingumyy, otherwise it's unknowable. Real random playlists don't exist, and would have repeats. pseudo random ones do exist, do have repeats and as they're machine generated to a set of unknown to us criteria without knowing those criteria we can't actually get the true answer(s). But it passed quite a bit of time.

I suppose the best answer we can come up with "it's fixed, like a FIFA draw", or soemething

[bickster]

But it does change the maths of the probability somewhat dependent on how many songs it downloads in advance, when it chooses them and how it then picks between those songs it has chosen.

All the maths above was based on each song being played next being a single event, my suspicions are that this isn't actually the case

[blandy]

5 minutes ago, bickster said:

each song being played next being a single event, my suspicions are that this isn't actually the case

Oh, sure. Definitely it isn't. I think I posted that somewhere as one of my assumptions.

[mjmooney]

Shuffle play on my phone once played consecutive songs (out of about 3,000 tracks) from solo albums by David Crosby, Stephen Stills, Graham Nash and Neil Young (although not in that order). 

[HanoiVillan]

AIUI, you can get truly random sequences of numbers (albeit short strings, and slowly) by using a Geiger counter. I guess you could also use random.org (which uses noise in the Earth's atmosphere) if you really wanted to spend the time programming the playlist. 

Edited June 28, 2018 by HanoiVillan

[Stevo985]

I'm sure I read once that squirrels had some kind of random movement generator built into their brains.

it's a defence mechanism that means predators can't read their movements.

A squirrel once fell off a tree and landed in the road in front of me and it did what I imagine is the output of that, and just seemed to go mental in the road, running all over the place (he survived)

 

However, now I can't find anything online to back that up. So I'm wondering if I made the whole thing up.

[tonyh29]

1 hour ago, Stevo985 said:

I'm sure I read once that squirrels had some kind of random movement generator built into their brains.

it's a defence mechanism that means predators can't read their movements.

A squirrel once fell off a tree and landed in the road in front of me and it did what I imagine is the output of that, and just seemed to go mental in the road, running all over the place (he survived)

 

However, now I can't find anything online to back that up. So I'm wondering if I made the whole thing up.

shame about the picture quality

[BOF]

1 hour ago, HanoiVillan said:

AIUI, you can get truly random sequences of numbers (albeit short strings, and slowly) by using a Geiger counter. I guess you could also use random.org (which uses noise in the Earth's atmosphere) if you really wanted to spend the time programming the playlist. 

TIL AIUI

[bobzy]

This thread does not deliver over the past couple of pages 

[Seat68]

Too much sums. 

[NurembergVillan]

On 27/06/2018 at 16:55, blandy said:

I don't think I agree, still, snowy.

If the playlist randomly plays 36 different tracks, and doesn't repeat any until the others have all been played then at any and all times, there are 35 different tracks to follow the currently playing track before we get to a possibility of a repeat. Once you hit the first track from Stevo's album, that's the only point you can then do the sums,  and you'd need to do then as per the method you initially outlined (but using the right figures  )

unless you want to do the really complex maths, which would basically be what is the chance of the first 4 tracks (of 36) being from the album (7/36 * 6/35 *  5/34 * 4/33), then factor in what is the odds of the first track not being from the album, but the next 4 all being from. so + (29/36 * 6/35 *  5/34 * 4/33 * 28/32 * 27/...etc.).. then then factor in what is the odds of the first two tracks not being from the album, but the next 4 all being from.+ (29/36 * 28/35 * 6/34 * 5/33 * 1/8 * 3/31 * 28/30 * 27/....etc.)   and so on. So that would work out the probability for 4 tracks from the album playing in succession at any point in a 36 track playlist).

We are miles away, but it's closer to going homw time and cricket watching

and also we don't have a set of defined assumptions, but that's the maths, as I understand things, at least.   

[HanoiVillan]

I'd be interested in people's answers to this:

 

[villa4europe]

im a QS and i would say just about everything...including by a lot of people within the construction industry and even within my profession, i've had several trainees who just dont get it

i control financial risks on a building site

 

[BOF]

5 minutes ago, HanoiVillan said:

I'd be interested in people's answers to this:

 

Logic.

[bickster]

14 minutes ago, HanoiVillan said:

I'd be interested in people's answers to this:

 

Seriously, my answer would definitely be TL:DR

But essentially @BOFis bang on the money

[lapal_fan]

33 minutes ago, HanoiVillan said:

I'd be interested in people's answers to this:

 

That energy isn't free. 

[Seat68]

That because I raise defects against shit code, that I am the problem for finding said defects.