• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

Jam88's Quest to be the next Feliks Zemdegs - Week 1 | Restarting, 4x4 PB

Lucas Garron

Administrator
Joined
Jul 6, 2007
Messages
3,718
Location
California
WCA
2006GARR01
YouTube
Visit Channel
@Lucas Garron Have you heard about the timer.cubing.net scrambles being easier than others from anyone/where else?

Not that I have heard.

timer.cubing.net is actually every careful to produce the best possible random-state scrambles where other timers (e.g. qqtimer, csTimer) currently do not. In fact, you might notice that the only currently supported events are ones where proper random-state scrambles are possible.

But it's always possible that there are bugs. I'm actually in the process of overhauling scramble generation right now (because I want all web timers to be able to use high-quality scrambles easily), so I'd be happy to review any evidence of incorrectly distributed scrambles.
 

Lucas Garron

Administrator
Joined
Jul 6, 2007
Messages
3,718
Location
California
WCA
2006GARR01
YouTube
Visit Channel
timer.cubing.net is actually every careful to produce the best possible random-state scrambles where other timers (e.g. qqtimer, csTimer) currently do not. In fact, you might notice that the only currently supported events are ones where proper random-state scrambles are possible.

I should probably clarify: timer.cubing.net, qqTimer, and csTimer use the same code for 3x3x3, which I adapted from Chen Shuang's Java code several years back. (Code for other puzzles is similarly shared.)

The only significant difference is that timer.cubing.net is using the browser's built-in high quality random number generation (crypto.getRandomValues() instead of Math.random()).
 

Jam88

Member
Joined
Mar 23, 2020
Messages
1,194
Location
inside a cube
YouTube
Visit Channel
I should probably clarify: timer.cubing.net, qqTimer, and csTimer use the same code for 3x3x3, which I adapted from Chen Shuang's Java code several years back. (Code for other puzzles is similarly shared.)

The only significant difference is that timer.cubing.net is using the browser's built-in high quality random number generation (crypto.getRandomValues() instead of Math.random()).
OK cool, thanks! Maybe the browser's built in thing needs updating
 

Lucas Garron

Administrator
Joined
Jul 6, 2007
Messages
3,718
Location
California
WCA
2006GARR01
YouTube
Visit Channel
OK cool, thanks! Maybe the browser's built in thing needs updating

Hmm, the browser API is almost certainly not an issue. Browsers know they need to be super careful with it, and it would be a very serious bug if the data was distinguishable from true randomness.
And if there was an issue, it's also unlikely that the issue would result in noticeably easier scrambles (for the same reason that qqTimer and csTimer are fine in practice).

I took a quick look at some basic scramble distributions (piece orientations, piece permutations), and they looked pretty normal. But I take scrambles very seriously, so I'll try to take a closer look.

Here's a JSON file of 100,000 scrambles if anyone else would like to run randomness tests on it:
 

Jam88

Member
Joined
Mar 23, 2020
Messages
1,194
Location
inside a cube
YouTube
Visit Channel
Hmm, the browser API is almost certainly not an issue. Browsers know they need to be super careful with it, and it would be a very serious bug if the data was distinguishable from true randomness.
And if there was an issue, it's also unlikely that the issue would result in noticeably easier scrambles (for the same reason that qqTimer and csTimer are fine in practice).

I took a quick look at some basic scramble distributions (piece orientations, piece permutations), and they looked pretty normal. But I take scrambles very seriously, so I'll try to take a closer look.

Here's a JSON file of 100,000 scrambles if anyone else would like to run randomness tests on it:
That's true. I just seemed to be getting a ton of free pairs and xcrosses etc
 

Jam88

Member
Joined
Mar 23, 2020
Messages
1,194
Location
inside a cube
YouTube
Visit Channel
Haven't really done any 3x3, been doing 9x9 :p
I hope to get back to 3x3 sometime in the next week.

I'm going to put my cubing life goals here.
Get the WR single and average in all nxn's and their variants (except BLD) and pyra, clock, skewb and squan.
Beat @Faz, Tymon Kolasinski, Max Park and Leo Borromeo in a head-to-head race.
Get 1 million subs.
Yes, I know I'm an underachiever
 
Top