Comments on: Crowdsourcing Code http://www.borismus.com/crowdsourcing-code/ Software Designer Tue, 27 Jul 2010 06:00:35 -0700 http://wordpress.org/?v=2.9.2 hourly 1 By: Boris Smus http://www.borismus.com/crowdsourcing-code/comment-page-1/#comment-278 Boris Smus Sun, 17 Jan 2010 16:49:36 +0000 http://www.borismus.com/?p=568#comment-278 It's pretty darn difficult to write any sort of reasonable unit tests for user interfaces, so I definitely agree with you there. But the interestingness of an application is determined by the designer, not a code monkeying turker, so I'm not sure what you mean. As far as profitability goes, perhaps this approach would allow a small number of software architects to roll out working (and unit tested) software very quickly, since turkers are easily parallelized. It’s pretty darn difficult to write any sort of reasonable unit tests for user interfaces, so I definitely agree with you there.

But the interestingness of an application is determined by the designer, not a code monkeying turker, so I’m not sure what you mean.

As far as profitability goes, perhaps this approach would allow a small number of software architects to roll out working (and unit tested) software very quickly, since turkers are easily parallelized.

]]> By: Allen Pike http://www.borismus.com/crowdsourcing-code/comment-page-1/#comment-277 Allen Pike Sun, 17 Jan 2010 01:28:11 +0000 http://www.borismus.com/?p=568#comment-277 The trick would be writing unit tests that were thorough enough in less time than it would take you to code it yourself. With that, you could add in Turks for QA and hey, maybe you could even get Turks to write the unit tests. I'm sure you could produce non-UI apps that function in basic tasks using this approach, but I doubt it could profitably extend to interesting or profitable types of apps. The trick would be writing unit tests that were thorough enough in less time than it would take you to code it yourself. With that, you could add in Turks for QA and hey, maybe you could even get Turks to write the unit tests. I’m sure you could produce non-UI apps that function in basic tasks using this approach, but I doubt it could profitably extend to interesting or profitable types of apps.

]]>