Don't like ads? PRO users don't see any ads ;-)
Public Pastes
Guest

Dev8D AU Gamification Minutes

By: a guest on Nov 7th, 2011  |  syntax: None  |  size: 8.68 KB  |  hits: 73  |  expires: Never
download  |  raw  |  embed  |  report abuse  |  print
Text below is selected. Please press Ctrl+C to copy to your clipboard. (⌘+C on Mac)
  1.  Hi Gamers!
  2.  
  3. People:
  4. Andrew Perry
  5. Ben Porebski
  6. Jared Winton
  7. Bosco Ho
  8. Ian Thomas
  9. Steve Androulakis
  10. Martin Paulo
  11. Tom Fifield
  12. Joanna Huang
  13.  
  14. taking gamelike elements and putting it into real world acitivites and make it more funa and engagin
  15.  
  16. there is fold-it. Published in nature biology.
  17.  
  18. the other end of the spectrum:
  19.  
  20. Australian archives of newspapers, editorial corrections
  21.  
  22. Crowdsourcing by librarians.
  23.  
  24. Crowdsourcing example: The UK politician 'receipts' scandal. The Guardian newspaper put together an app really quickly and challenged readership to find misuse of public funds amongst thousands upon thousands of records.
  25.  
  26. A need for 'gaming' not being exploited enough right now.. Launch Pad.. translating pieces of language. There's a 'game' system but it's not upfront or prominent enough
  27.  
  28. Game: Galaxy Zoo. Classify galaxies from pictures, via a simple classification scheme (crowd sourced). - http://www.galaxyzoo.org/how_to_take_part
  29.  
  30. Andrew Perry's Android game. Like hot or not. Called Grandmaster Pixel. Rate user contributed pixel art by choosing the best of 2 candidates at a time. As you submit more pictures, you get more colours in your palette. - http://www.grandmasterpixel.com/ http://omgwtfgames.com/
  31.  
  32. http://130.194.155.72/
  33. andrew.perry
  34. Rigaku CrystalTrak software shows crystal plates and lets 'crowds' describe them ie 'Clear' 'Clear with stuff', Good/Bad Precipitate
  35.  
  36. Does this technique work outside of a few options? Perhaps too many options of classification = inaccurate results
  37.  
  38. Perhaps the practice of crystal screening can't be compelling enough to hold wider attention outside of those with a personal interest?
  39.  
  40. Does it make more sense to expand this focus of this kind of classification application to general "science images"? Rather than domain-specific images.
  41.  
  42. Random sub-sections for "Scientific Imaging: The game". Might have more chance of enticing the wider public.
  43.  
  44. Bring more image classification candidates to someone established, like Galaxy Zoo and use their hardware.
  45.  
  46. Leaf-minor moth been killing leaf-minor trees. Look at a series of leafs and rate them on damage. There's an issue with a high drop-off rate. Ie lack of persistent community. Galaxy Zoo has the benefit of being attractive to users ie. "look at the pretty galaxies". One way to combat this is to give the ability to export this data in a usable format.
  47.  
  48. The Fold-it game example: Game is fun, interface is good, haven't discovered anything fundamental yet. Perhaps the benefits are minor. Massive publicity though.
  49.  
  50. Baker lab: Fold-it @ home too.. much like the seti idea.
  51.  
  52. *********
  53.  
  54. THIS AREA IS ABOUT SOFTWARE TRENDS:
  55.  
  56. why are you shouting? LOUD NOISES
  57.  
  58. People:
  59. Andrew Perry
  60. Ben O/Steen
  61. Ben Porebski
  62. Bosco Ho
  63. Ian Thomas
  64. Steve Androulakis
  65. Martin Paulo
  66. Tom Fifield
  67. Joanna Huang
  68. Shaun O'Keefe
  69. Paul Walk
  70. Richard Rothwell
  71.  
  72. Martin Paulo sees a big shift in software... (death of flash) HTML5, mobile devices, touch, javascript and html taking an almost ubiqutous stake.
  73. Virtualisation one physical piece of hardware to run many virtual machines that could be anywhere. Cloud infrastructure or software or platforms as a service.
  74. Applications being able to run online and offline (perhaps another offshoot of HTML5).
  75. Agile movement is the dominant method.
  76. Continuous deployment of web sites.
  77. Constant expectations for app updates changing the old 'pay for major updates'.
  78. Advertising driving apps, not subscriptions or fees.
  79.  
  80. What makes a good developer in this ecosystem? This has changed much from what's taught in comp sci.
  81.  
  82. Comp sci is always a generation behind.
  83.  
  84. There's a notion in the UK about 'coding for kids'.
  85.  
  86. Paul Walk ran software teams and most of what needed to be taught was breaking out of the traditional learned practices at uni: ie. No UML before development, working together as a team in an agile manner.
  87.  
  88. security - is web security enough? SQL injection attacks etc are not necessarily covered in the basic courses, but serious concerns!
  89.  
  90. Web development.. a tool in your toolbox must be web security.
  91.  
  92. That kind of knowledge needs to be in infrastructure from the get-go
  93.  
  94. Two separate ideals: Software scaling etc often opposes "just get it done methodology"
  95.  
  96. International particle physics project scaling problem: Latency across countries. In this area, a timeout of 1 second could grind an application to a halt.
  97.  
  98. The risk that some critical service would 'disappear'. Ie, go down due to constraints of scale.
  99.  
  100. Such a thing as a 'burn line'. Whatever is above it is what parts of a system you can lose - below are the critical components. So service levels within an app and knowing what features will take what beating.
  101.  
  102. At CERN there's a 100 point system for levels of needed service. 100 being okay to lose for a while. 0-20 must be available within minutes, for example.
  103.  
  104. Many rich javascript applications are ripe for attack and their developers aren't neccissarily those with a focus on security
  105.  
  106. SSL isn't the magic bullet here. 'web of trust' is a bit of a lie..
  107.  
  108. Latest version of ruby on rails has deprecated the standard SSL plugin.. works as a switch for entire application. Previously you'd leverage it for parts of apps (controllers).
  109.  
  110. Is there a case for 'peer programming' for security. "Try and break my software".
  111.  
  112. Perhaps Australia being a smaller size to other major groupings (NA/Europe), surely we can fund a national infrastructure that will hope to make environments more secure. (see also: NeCTAR)
  113.  
  114. Microsoft going from their propietry systems (Visual Basic, the best selling language fo all time and .NET stable) to HTML/javascript 'open' based tech for Windows 8.
  115.  
  116. In the old days, what you trigger in an interface would cause a reaction. These days of the web and asynchronous services... what you may trigger may never result in an action. What to do then? Cant guarantee any level of interactivity.
  117.  
  118. Current programaing paradigms largely don't address coding for multi-core or multi-cpu/gpu systems. It requires specialist techniques and frameworks.
  119.  
  120. Are we going to see grant bodies assessing whether researchers are able to parallelise work computationally for maximum affect? Could mean a lot in real terms.
  121.  
  122. Current practice of putting in grants for things that aren't possible or financially feasable at the time and 'hopefully' or will 'likely' be feasable toward project's end.
  123.  
  124. In the US, universities own the IP of the software.
  125.  
  126. At Melbourne Uni at least, the researcher owns the IP.
  127.  
  128. Is software patentable in the first place?
  129.  
  130. Nothing wrong with paying fees for software? So long as price is right. Raspberry pi, software company providing a small device with a series of ports.. very powerful chip, capable of doing video processing, very cheap. 5-10% of fees are licensing cost. For video encoding, 20c per chip. ARM: 1c. Dolby: AAC audio want $1 per chip.
  131.  
  132. Is the dolby price fair as its likely a patented mathematical algorithm?
  133.  
  134. Can Agile exist across distributed networks of people on projects? Agile falls down as a distributed team aiming to solve a research problem: people go into many different directions. Though it can work as long as you write down what's being done. Central task list and people would pick what to work on. Worked across 14 countries (Tom Fifield's team). A week-long hackathon many times a year together. Anyone could put tasks onto the list, with a guiding hand defining them. Given there was so much work, it was fairly effective in this ad-hoc way. Collected experts in specific areas in sub-teams. So perhaps many agile projects, not a single? Though it ran as one. Didn't always work, but mostly a success. Not even a test structure. This project would deploy onto test infrastructure when new features would be 'completed'. A python/c++ component. C++ because there were some interactions with other C++ modules. Swig to wrap c++ in python across systems. Doesn't work that well for many applications.. Proved a major challenge in this project. Swig is older than some but has been somewhat of a mainstay.
  135.  
  136. Use IRC and mailing lists to combat time zone changes with teams doing the same work.
  137.  
  138. There are vicous Agile zealots out there. Some of us distance ourselves from this. Agile to us is more small-a 'agile'.
  139.  
  140. Problem with Agile in a research environment is getting a hold of researchers. Researchers are busy. Put software developers in the lab with the researchers.
  141.  
  142. Working in science as a software developer can be difficult though as many scientists believe they can be software developers without the formal training and often their bad practice can cause their software to collapse in on itself.
  143.  
  144. There's a really good PDF on python performance. Covers pypy, syphon, pycuda, parallel python. bit.ly/uIu65f
  145.  
create a new version of this paste RAW Paste Data