Difference between revisions of "Milestone 8 TuringMachine Foundation1"

From crowdresearch
Jump to: navigation, search
(Challenges & Solution)
 
(2 intermediate revisions by one other user not shown)
Line 6: Line 6:
 
The main idea is to maintain the submission approach from microtask markets, which is focused on two to hundreds of replications, but find ways to make it accessible to both microtask and expert work.
 
The main idea is to maintain the submission approach from microtask markets, which is focused on two to hundreds of replications, but find ways to make it accessible to both microtask and expert work.
 
   
 
   
As Adam Marcus mentioned last week: "Why would a programmer want to join a crowd work platform? The draw of AMT right now is that you can find jobs quickly. Can we enable that same benefit for programmers?”
 
 
 
====Challenges & Solution ====
 
====Challenges & Solution ====
 
*'''Now we need to figure out: what would such a marketplace look like? Is there a way to adapt a microtasking model so it feels natural and useful for macrotasks? '''<br>
 
*'''Now we need to figure out: what would such a marketplace look like? Is there a way to adapt a microtasking model so it feels natural and useful for macrotasks? '''<br>
 +
**
  
  
Line 24: Line 23:
  
 
*'''Why would a programmer want to join a crowd work platform? The draw of AMT right now is that you can find jobs quickly. Can we enable that same benefit for programmers?'''<br>
 
*'''Why would a programmer want to join a crowd work platform? The draw of AMT right now is that you can find jobs quickly. Can we enable that same benefit for programmers?'''<br>
 +
**
  
  
Line 42: Line 42:
  
 
*'''What would the tasks look like? How are they submitted? Does this look like AMT where any expert certified in an area can accept the task? Or like an oDesk negotiation?'''<br>
 
*'''What would the tasks look like? How are they submitted? Does this look like AMT where any expert certified in an area can accept the task? Or like an oDesk negotiation?'''<br>
 +
**
  
  
Line 55: Line 56:
  
 
*'''How do we ensure high-quality results? Do you let an expert work for hours and submit? That seems risky. Should there be intermediate feedback mechanisms?
'''<br>
 
*'''How do we ensure high-quality results? Do you let an expert work for hours and submit? That seems risky. Should there be intermediate feedback mechanisms?
'''<br>
 +
**
  
  
Line 64: Line 66:
  
 
*'''How do you trust that someone is an expert?
'''<br>
 
*'''How do you trust that someone is an expert?
'''<br>
 +
**A small task similar to the job to be completed could be provided by the requester to the worker. This could test whether the worker actually has the skill set required to complete the task assigned by the requester.

Latest revision as of 00:49, 23 April 2015

Foundation 1: Micro+macrotask market

Create a marketplace that spans micro- and macro-tasks. Could the same marketplace scale from 2 to N people not just labeling images, but also Photoshopping my vacation photos or mixing my new song?

The main idea is to maintain the submission approach from microtask markets, which is focused on two to hundreds of replications, but find ways to make it accessible to both microtask and expert work.

Challenges & Solution

  • Now we need to figure out: what would such a marketplace look like? Is there a way to adapt a microtasking model so it feels natural and useful for macrotasks?







  • Why would a programmer want to join a crowd work platform? The draw of AMT right now is that you can find jobs quickly. Can we enable that same benefit for programmers?









  • What would the tasks look like? How are they submitted? Does this look like AMT where any expert certified in an area can accept the task? Or like an oDesk negotiation?







  • How do we ensure high-quality results? Do you let an expert work for hours and submit? That seems risky. Should there be intermediate feedback mechanisms?






  • How do you trust that someone is an expert?

    • A small task similar to the job to be completed could be provided by the requester to the worker. This could test whether the worker actually has the skill set required to complete the task assigned by the requester.