09/07 - It's Wednesday, right?

[Jaded]

Related and interesting: https://medium.com/@dbreunig/do-alg...n-or-cause-depression-2e047ef84cda#.ta06dq3hu

Working on Mechanical Turk, it appears, is crushing.

For real,

Spoiler

Log in or register now. to view Spoiler content!

 

[Kerek]

For real,

Spoiler

Log in or register now. to view Spoiler content!

At least I have money now.

 

[Jharkan]

..thank you..

you're welcome

 

[electrolyte]

I like how they attack someone else's work as laughable and unsound then go on to make the mistake of assuming mturk causes depression rather than considering the possibility that depressed people choose to work on mturk.

Yeah. The part I thought was particularly interesting was that if a ton of MTurk is depressed (whether before MTurk or because of MTurk), does that change how algorithms and machines are being trained and how will that affect how the algorithms function at the end?

 

[Blue]

I like how they attack someone else's work as laughable and unsound then go on to make the mistake of assuming mturk causes depression rather than considering the possibility that depressed people choose to work on mturk.

The incomplete story is the most compelling though. Just think of that recent reporter who ONLY wanted to talk to turkers in California.

The guy raises some valid points though... but it's a really short and incomplete article.

 

[Kerek]

Yeah. The part I thought was particularly interesting was that if a ton of MTurk is depressed (whether before MTurk or because of MTurk), does that change how algorithms and machines are being trained and how will that affect how the algorithms function at the end?

I've always though that these studies they're doing are all biased by the self-selection of mturk workers, but supposedly they have the correspondence to known responses on the common questions, so I don't know. There are probably some quirks in the community that are going to affect the more subjective stuff.

 

[humbleturker]

I'm happier now than I've been in years now that I'm working (and surviving) online.

 

[coffeeowl]

At least I have money now.

 

[T. Leela]

guys
i have a crush on a girl I work with
a crush
i'm 35 years old and I'm using a term that I've always designated as "teenage use only"
i'm so pathetic

ps youre not pathetic.

 

[electrolyte]

I've always though that these studies they're doing are all biased by the self-selection of mturk workers, but supposedly they have the correspondence to known responses on the common questions, so I don't know. There are probably some quirks in the community that are going to affect the more subjective stuff.

Oh, there's tons of self selection. There was an article in the last month or so about how the drop out rate of workers midway through a survey on MTurk is HUGE compared to a lab (duh), and how that affects who is actually completing it.

 

[Kerek]

You're right. If I hadn't had this as an alternative I might have actually put my skills to work and built a career.

 

[Randomacts]

I'm happier now than I've been in years now that I'm working (and surviving) online.

mturk is actually the first and only job I've ever done.. it makes me happy actually to do it.. I just know I need more money then even the best of mturk can do to live here in CA

 

[SerJorahsFeetL1nt]

Title: Take a 5-minute survey with 70-cent bonus opportunity!! | PANDA Requester: cmu study [ALQ8DTU4XFXQ] (TO) TO Ratings: ★★★★★ 3.29 Communicativity ★★★★★ 4.24 Generosity ★★★★★ 4.50 Fairness ★★★★★ 4.61 Promptness Number of Reviews: 33 | TOS Flags: 0 Submit a new TO review Description: Participate in a Carnegie Mellon University research study. Time: 3 hours HITs Available: 1 Reward: $0.55 Qualifications: HIT approval rate (%) is greater than 95;Location is US

 

[Jesicu]

Oh, there's tons of self selection. There was an article in the last month or so about how the drop out rate of workers midway through a survey on MTurk is HUGE compared to a lab (duh), and how that affects who is actually completing it.

If they didn't hide writing in the middle, that number would be a lot lower.

 

[loki3404]

Yeah. The part I thought was particularly interesting was that if a ton of MTurk is depressed (whether before MTurk or because of MTurk), does that change how algorithms and machines are being trained and how will that affect how the algorithms function at the end?

That explains Marvin....

 

[Kerek]

Oh, there's tons of self selection. There was an article in the last month or so about how the drop out rate of workers midway through a survey on MTurk is HUGE compared to a lab (duh), and how that affects who is actually completing it.

Oh, and of course what they're paying is a huge selector as well. (Related to the dropout I'm sure)

 

[Blue]

If they didn't hide writing in the middle, that number would be a lot lower.

That's pretty simplistic... there are LOTS of reasons why people nope out of surveys.

 

[Bluedogan]

Tried one Frank. No. Oh boy no thank you

Are you all talking about the ones where you find the companies?

 

[Pleco]

Yeah. The part I thought was particularly interesting was that if a ton of MTurk is depressed (whether before MTurk or because of MTurk), does that change how algorithms and machines are being trained and how will that affect how the algorithms function at the end?

That's a comforting thought that through our work we maybe be influencing algorithms to be super depressed and self-loathing. Computers should be miserable too.

 

[Kerek]

That's pretty simplistic... there are LOTS of reasons why people nope out of surveys.