Summer Milestone 5 MichaelBernstein MicroMacro
I took an old Mechanical Turk task that my group had used. It involved using crowds to verify how often certain situations happen:
It's kind of long and arcane. I asked three workers to do it for 12 different inputs (a small subset of the entire task). At the bottom I asked workers to provide feedback on the task.
Here is some of the feedback they gave me:
- "Some actions are not very clear."
- "I think it needs to be less vague."
- "fix how the grammar lines up, add the ability for workers to write in their own options"
- "The actions don't fit the sentence very well."
- "there are many different drop down menus that are not used that take up much of the space in the hit. You could make them less long so that you could put them side by side."
- "Better instructions on what is needed to be done. More examples will help as well."
- "It seems OK. The "No action" entries are a bit weird, but not really a problem."
Overall, these seem like reasonable suggestions about how to improve the HIT, though not groundbreaking in how to redesign it. (I wouldn't have expected that.)
The results suggested ways in which workers made errors relative to what I would have wanted to see. I think this raises the question --- what should I do when there are errors? Redesign the HIT? Talk to the workers?
Next, I posted a job on Upwork: Sketch a logo for our Stanford research group This is a "phase 0" of our logo design project. We'll hire a few different designers for this sketch, pay you a fixed price for this phase, and then take the best logo idea and hire that person for a longer contract to make the logo digitally. ---
We are the "Stanford HCI Group". Our research group at Stanford is looking to create a new logo for our web page. Our research designs technology that makes people smarter, more powerful, or more fluid in their everyday lives. We involve a lot of "design thinking" (google IDEO to get an idea).
An inspirational website would be the Stanford d.school: http://dschool.stanford.edu
Specs: we're looking for something wide and short, like the d.school logo above. 16:9 aspect ratio would be fine, or anything somewhat similar. It can be a combination of text and abstract graphics.
--- Deliverable: at least one logo idea for our research group. These can be pen/pencil sketches, captured via a phone camera or similar. They don't need to be refined, we're mainly looking for concepts right now.
I went with three hires --- two Upwork recommended workers, and one not recommended. I can see why she wasn't super highly recommended. I got back two of the three in time.
One from the newbie worker:
Two from one of the recommended ("high level") workers: