[Tux4kids-discuss] [GSoC] Tux Racing UI feedback

Sat Jul 28 03:08:11 UTC 2012

Good evening!

On 7/26/2012 11:19 PM, Brendan Luchen wrote:
> Hey Kento,
>
> Kento Locatelli wrote:
>> Hi Brendan,
>>
>> I agree, displaying the stars/progress wasn't well thought out, it was
>> only intended to verify the problem generators were working properly.
>> That said, I'll probably just change the star display to be fractional
>> instead of integer. Maybe rounded to 1/4 star for display on the
>> exercise page, so it's obvious when an exercise isn't at 100%.
>> However now that I've started working on the game play, I'm not to sure
>> about the whole "stars ~ correct answers" idea... originally I intended
>> on having problem solving rate be proportional to player velocity, but
>> in that scenario it's logical to tie stars to finishing position (rather
>> than % correct). That system brings up other issues though, including:
>> ensure player is answering fairly and not quickly skipping
>> hard/answering easy questions, having the NPCs compete fairly, and
>> making stars explicitly represent speed but implicitly accuracy so the
>> game is still about racing while requiring high correct % to finish
>> first. NPCs is simple if problems specify finishing times for 1st/2nd
>> position. Exercise can also specify how long the game is, how much
>> energy correct answers provide, and incorrect answers zero the energy
>> meter and have a cool-down time to prevent quickly the first issue.
>> Energy steadily depletes and the ship slows as the energy meter goes
>> down, so the player must continuously answer questions correctly. In
>> practice I'm not sure how well that solves the 3rd issue though... does
>> anyone else believe 100% correct answers should be required for 1st 
>> place?
>
> Speed vs. accuracy, always a fun one. I can see two main approaches: 
> treating them as totally independent and orthogonal, or weighting 
> accuracy against speed by some factor (this could perhaps be set by 
> instructors).
>
> In a more general sense, we have a solid, generic question backend and 
> a basic space race aesthetic. We need to nail down the specific 
> gameplay mechanics that will lie on top if it. Brainstorming time! 
> Catch you on IRC.
What we discussed on IRC ended up being too complex, so I simplified and 
came up with something based on weighting accuracy/speed :
- Exercise set minimum correct answer rate for 1st / 2nd place
- NPCs randomly assigned one of those rates
- NPC distance to player is proportional to log(NPC rate - player rate) 
* random factor. This keeps them close so the race feels more competitive

- Player ship velocity is proportional to the correct answer rate * 
percent correct
- Rate and percent are over the entire game, so finishing position 
represents aggregate performance
- Proportion specified by exercise, so harder exercises still have fast 
races
- Time spent on incorrect answers is ignored in calculating rate, since 
the correct answer has to be displayed and that would impact exercises 
differently based on difficulty.
- Rate is displayed as the energy bar, percent is displayed via damage bar

- Exercise specifies time limit, infinite number of questions. So 
basically the player has to maintain a certain rate/accuracy for the 
duration to get a certain position

Here's a sample graph for position @ http://bit.ly/N2G2 (note: WebGL) It 
plots 0...1 accuracy on the y-axis and 0...10 problem/minute on the 
x-axis. 1st has a minimum weighted rate of 9, 2nd 7.
>
>>
>>> Some minor UI things I noticed
>>> - "Settings and Quit" doesn't need to be a submenu.
>> I'm not sure what you mean? On Android it would make sense to hide the
>> link since we have the menu button, but I'm not sure how to avoid the
>> settings/quit menu altogether.
>
> I think I accidentally a sentence or two. I meant that typically 
> "Quit" is its own option on the main menu, and doesn't fall under 
> "Settings". The two options aren't totally necessary on Android, no, 
> but it's nice to have if there's room. Then there's the fact that 
> Settings only has one child, but I'm sure that will change :)
Oh I see, that does look awkward... Split!
>>
>>> - Looks weird in landscape. Probably the CSS needs some massaging, or
>>> we can just force a portrait orientation, which isn't uncommon for 
>>> games.
>> Which resolution are you using? For my devices portrait is the awkward
>> one, with tiny icons / text and large white space.
>
> Right you are. I had it backwards. It's been a long week.
>
>>
>>> - No indication of the correct answer (Pere mentioned already)
>> The lines didn't show up very well on the sketch scan but the lower
>> portion of the screen dedicated to answers will flash green and chime on
>> correct, and red + error beep on incorrect.
>
> But will it tell you what the right answer was if you mess up?
Yes, probably by outlining the correct answer's box.
>
>>>
>>> Looks good on the whole.
>>>
>>>
>>> Cheers,
>>> Brendan
>>
>> Thanks :D
>> - Kento
>>
>>> Kento Locatelli wrote:
>>>> Thank you for trying it out!
>>>>
>>>> On 7/22/2012 2:46 AM, Pere Pujal i Carabantes wrote:
>>>>> El dv 20 de 07 de 2012 a les 00:00 -0700, en/na Kento Locatelli va
>>>>> escriure:
>>>>>> Hi everyone!
>>>>>>
>>>>>> The Tux Racing cross-platform/mobile game project is about 3/4
>>>>>> done, and
>>>>>> I would like your assistance in testing what is complete so far.
>>>>>>
>>>>>> Thus far most of the work has been with the "infrastructure" part
>>>>>> of the
>>>>>> app so it's not particularly interesting yet, but the rest of the 
>>>>>> game
>>>>>> is based on this so I need your help in verifying it works
>>>>>> properly! So
>>>>>> please mess around with it a little and let me know of anything that
>>>>>> comes up, whether it's a bug or suggestions for improving the
>>>>>> aesthetics. I am especially interested in problems with the problem
>>>>>> generator or UI on desktop browsers/iOS/Android, although 
>>>>>> feedback on
>>>>>> other problems/smartphone platforms is welcome as well. Relatively
>>>>>> recent Firefox/Chrome/Chromium serve as a good reference.
>>>>>>
>>>>>> Android APK (Android 2.2+):
>>>>>> https://github.com/klocatelli/TuxRacing/TuxRacing.apk/qr_code (May
>>>>>> have
>>>>>> to temporarily allow non-Marketplace apps)
>>>>>> Web-based: http://www.klocatelli.name/TuxRacing/#/index (After 
>>>>>> loading
>>>>>> once, should work offline in relatively modern browsers)
>>>>>> iOS package will be released with the final version, but currently
>>>>>> pending response from Apple regarding beta releases
>>>>> Testing in a hurry
>>>>> Only testing on iceweassel 10.0, don't have android:
>>>>> language settings ca es en : error no modules for your locale, should
>>>>> default to english?
>>>>> language settings en ca es : runs
>>>> Unfortunately the game only has access to the highest priority 
>>>> language
>>>> setting, so it doesn't know you have English enabled at a lower
>>>> priority. But I agree, it's a bit odd to have the game default to a
>>>> different language (English) and then tell you there aren't any 
>>>> modules
>>>> for your language... I changed it so everything defaults to 
>>>> English, and
>>>> lessons that support your language show up at the top of the lesson 
>>>> list
>>>> while lessons that don't show up lower down with a warning that 
>>>> they're
>>>> in another language.
>>>>
>>>>> playing chemistry:
>>>>> significant figures: Don't know what are significant figures so 
>>>>> figure
>>>>> kids...
>>>>> density: haven't managed to get recognized right responses.
>>>> Turns out the density problem was a bug in how answers were checked in
>>>> the chemistry lesson, it should work as expected now.
>>>> I agree that the questions don't fit the purpose of the game, but they
>>>> are for demonstration purposes only :)
>>>>
>>>>> playing arithmetic: right responses are recognized, Perhaps one would
>>>>> expect a map 1<->1 from right responses to stars. Putting the 
>>>>> questions
>>>>> in small blocks of 5 questions each is great.
>>>> Isn't that what it's doing right now? Every question is worth 3 / 
>>>> total
>>>> number of questions. Right now all exercises are 5 questions long, so
>>>> every question is worth 0.6 stars. If an exercise was 15 questions 
>>>> long,
>>>> every question would be worth 0.2 stars. The stars count is rounded
>>>> though, so if you get 2 questions right you only see Stars = 1 (even
>>>> though it's 1.2).
>>>>
>>>>> The ok dialog is a pain, it should autovanish and show the right
>>>>> response in case of error, we are teaching after all.
>>>>> The number of stars and the title of the exercise perhaps should
>>>>> show on
>>>>> the top of the question
>>>> I agree, the current exercise page is a pain... however it's just for
>>>> demonstration :D My previous email should have had some attachments
>>>> showing what the final exercise page will look like, though I will 
>>>> alter
>>>> it to include your suggestion of the exercise title / stars.
>>>>
>>>>> No mute/unmute here
>>>>>
>>>>> HTH
>>>>> Pere
>>>> Thanks, that was another bug! Should be fixed now, though I tested 
>>>> with
>>>> Iceweasel 11 instead of 10
>>>>
>>>> Cheers!
>>>> - Kento
>>>>
>>>> _______________________________________________
>>>> Tux4kids-discuss mailing list
>>>> Tux4kids-discuss at lists.alioth.debian.org
>>>> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/tux4kids-discuss 
>>>>
>>
>>
>> _______________________________________________
>> Tux4kids-discuss mailing list
>> Tux4kids-discuss at lists.alioth.debian.org
>> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/tux4kids-discuss