1.8.9.1 BubblePop bubbles should not differ in size on the same screen.

[JackMostow]

Sharon Carver critiqued our use of BubblePop for numeracy tasks:
… The randomly chosen color for each bubble is probably harmless, but the randomly chosen size is distracting and, at least while we were watching, the larger number more frequently had the larger size bubble, so children may be trying to use the size of the bubble to make the judgement and be confused. Children have a concept of bigger and smaller that has to so with size, so you shouldn’t have size be a variable when you are trying to choose greater than / less than. That’ll just confuse.

Bubble size (height and width?) presumably gets determined and randomized somewhere in RoboTutor/comp_bubblepop/src/main/java/cmu/xprize/bp_component/ bubble.

[JackMostow]

From: Nirmal Patel nirmal@playpowerlabs.com
Sent: Friday, June 22, 2018 1:21 PM
To: Jack Mostow mostow@cs.cmu.edu
Cc: Derek Lomas derek@playpowerlabs.com; Kevin Deland kdeland@andrew.cmu.edu; Amy Ogan aeo@andrew.cmu.edu
Subject: Re: why vary bubble size in BubblePop?

No problem Jack. I can stay involved in the issue.

[JackMostow]

@nirmalpatel - When possible please reply via the relevant GitHub issue to keep them sorted and avoid runaway nesting.

I invited you as a RoboTutor repo collaborator with write privs so you can edit the relevant file(s). Please follow the protocol summarized in item 10 of README -- Information for RoboTutor team members and summarized at .

Thanks! - Jack

[JackMostow]

@nirmalpatel - Thanks for agreeing to diagnose and fix the within-screen bubble size variation problem.
I assume it's somewhere in RoboTutor/comp_bubblepop/src/main/java/cmu/xprize/bp_component.
Do you have the information needed to track it down, fix it in a new branch, and test the fix?

1. I initially assumed that bubble size is represented by the height and width variables. But there's also a scale variable.

2. I also assumed that the problem is caused by a deliberate randomization step somewhere, so I was looking for "rand". But perhaps it's an unintended side effect of some non-random decision, such as altering bubble size to prevent overlap when Derek Tam (@dptam) extended BubblePop to display bubbles large enough to display more than one character or line of text.

3. To narrow down when the change I occurred, I looked at some old screen videos in FROM TABLETS > From non-ASR GLASS 2016-06-10 (screen video has audio):

2016_06_22_12_33_38.mp4: same size

2016_11_12_12_58_12.mp4: same size

2016_11_24_17_18_38.mp4: same size [last video of 2016]

SO THE CHANGE HAPPENED NEAR OR DURING 2016 WINTER BREAK

2017_01_02_15_39_19.mp4: size varies [first video of 2017]

2017_03_21_17_27_01.mp4: size varies

Is there a way to search for commits during a specified date range?

4. Promisingly named branches include:

• BubblePopmod
• bp-resize (!)
• dptam
• dptam-patch

[nirmalpatel]

@JackMostow @kevindeland I think I have spotted the issue. Seems like it is very simple to fix it, so I'm not pushing it to the repo.

The problem IS caused by deliberate randomization steps in 4 different places.

In https://github.com/RoboTutorLLC/RoboTutor/blob/master/comp_bubblepop/src/main/java/cmu/xprize/bp_component/CBp_Mechanic_MC.java line 419, it says:

SBubbles[i1].setAssignedScale(getRandInRange(_scaleRange));

This needs to be changed to:

SBubbles[i1].setAssignedScale(1.0f);

In https://github.com/RoboTutorLLC/RoboTutor/blob/master/comp_bubblepop/src/main/java/cmu/xprize/bp_component/CBp_Mechanic_RISE.java lines 222, 241, and 264, it says:

nextBubble.setScale(getRandInRange(_scaleRange));

This can either be commented, or changed to:

nextBubble.setScale(1.0f);

Doing this on my end gives me following view in the RT where it seems all the bubbles are of the same size:

Please let me know if you still need me to make changes and push. The main reason I'm not doing it is that development branch is not working out for me (it starts by default with a tutor with no way to go back to the activity selection screen.) I still have created a branch out of the master branch to make these changes, though.

-- Nirmal

[JackMostow]

Nirmal - Great! Please configure to generate a debug apk rather than a release apk. If RoboTutor input a separate configuration file, we'd only need a single apk.

Nice sleuthing -- I especially appreciate your thoroughness in finding all 4 places, not just one. - Jack

[JackMostow]

@nirmalpatel - Please push this change if you didn't already, so we can close the issues. Thanks! - Jack

[kevindeland]

merged