downloading folders from google drive.

I wanted to download some course material on RL shared by the author via Google drive using the command line.  I got a bunch of stuff using wget a folder in google drive was a challenge. I looked it up in SO which gave me a hint but no solution. I installed gdown using pip and then used: gdown --folder --continue if there are more than 50 files you need to use --remaining-ok and only get the first 50. In such a case its best to download using the folder using the UI and decompress locally. Decompressing from the command line created errors related to unicode but using the mac UI I decompressed without a glitch.

Effects of cognitive surplus on some of all the knowledge

MIND THE GAP (Photo credit: sarahstierch)

At Wikisym last year I discussed an unpopular opinion that readers who see errors and don't fix them are free riders and how a similar case holds for readers who do neither edit nor contribute to the fundraiser. Of course there are many caveats such as not ever seeing the edit button, difficulties in using the UI and of course the community.

 At the time the feedback tool had been launched and this had lowered the bar to participation. Until the community decided other wise - raising the bar to participation and cynically pointing out that in most cases this was worse then free riding - it was encouraging "parasitic" behaviour - participation that was not constructive and which only sucked up resources. As my own experience in the Yoga project was radically different I had not bothered to go into this.
One of the questions that tends to come up when considering women and the gender gap is in Wikipedia is:

Are the 84% of women who don't edit Wikipedia free riders?

Of course that number is bogus scientifically but the question of free riding and its proportions does remain particularly when considering women's contributions. Though it was argued in  a CSCW paper that readers are not free riders. But subsequent abuse of the feedback tool seems to have laid that rather unconvincing theory to rest.

This is a question that I acknowledge internally but then move on. It would clarify things if it could tested and rejected. But this is not so simple. How do you even measure non participation of this type. Survival analysis does not seem to be of any use, matching and latent variables are not directly applicable either.

The only viable method I can see is to simulate using game theory. The most closely related classic game being the stag hunt. But even in this case there are difficulties stemming from defining a suitable payoffs to define suitable incentives of free riders. 

All this leads me to consider alternative, more convincing models for non participation. Borrowing some mind experiments from Toffler and McLuhan I can check if these are at least more convincing intuitively before looking for qualitative evidence in existing CSCW research and trying to build mathematical model.

Hyperbolic discounting meets Cognitive surplus.

Since writing a decent bio is within reach of most beginning editors why are women not doing this? Outreach at WM.IL teaches a very comprehensive course on Wikipedia in about 3 hours.

Today even writing a new article does have three or more gatekeepers.
  1. Need to register
  2. Need to know to make a 
    1. Main space draft - risky
    2. Userspace draft
    3. AfC draft this introduces a delay
  3. Get someone at recent change patrol to cooperate and approve the draft. (This introduces a delay penalty)

If we consider that information/time is correlate of productivity for Wikipedia editor and also a correlate of the fun factor we can begin to understand lower participation.

But in the fun economy we have agents perusing self-interest . How rational are they for not editing Wikipdia or for inventing rocket-brain surgery for that matter. Lets consider that time is of the essence and introduce a suitable utility function for free time which would show that women are more rational than men.

The quasi-hyperbolic cognitive surplus utility (QHCSU)

$$U(t) = u + b x \delta^t  \space (quasi-hyperbolic discounting)$$

u is a static utility , b and delta are parameters between 0 and 1.
b corresponds to a discounting for non immediate gratification and 
delta represents the discounting of future events in subsequent time periods. 

A person has the following options online :

Google       b~ .1 sec.   u =  1 min
Instagram    b~  1 sec.   u =  4 hr  * friends
Twitter      b~  5 sec.   u =  1 hr * friends
Facebook     b~ 30 sec.   u =  4 hr * friends
Wiki(read)   b~ 20 sec.   u = 20 min
Blog         b~ 20 min.   u =  3 days * readers
Wiki(edit)   b~ 4.5 hr.   u =  1 months / enemies
Hack floss   b~ 999 hr.   u =  1 year * users / 1,000,000

Women tend to keep better track of social relations, so will have more friends than men
and there is delta as well - it is almost certainly different for men and women.
social relations: 
  • enemies on Wikipedia  are highly persistent.
  • friends are weak ties.
  • readers are weaker but hardly a social one.
  • users are a motivation but not a social one.

Women have less free time

A common reason when excusing themselves from editing Wikipedia is that women have less free time, particularly when they raise children. Certainly, since this is the main reason they work less and get paid less when doing the same work. But they will still post pictures to Instagram, Tweet and use Facebook. So even is women have less free time the real issue is that they prioritise differently.

Looking at the above utility and just ignoring delta, women who are general more social have more friends and so will get less utility from reading or editing Wikipedia.

Putting in different deltas for men and women will make the difference even more pronounced.

Further work

Further work in this area involves:
  • Running an experiment to measure hyperbolic discounting for wiki and social media.
  • Using a simplified QHCSU in games modelling the gender gap to create more complex behaviour.
The experiment is on hold since certain vital statistics become available from the WMF analytics team. Particularly how much time/words do new and older editors add on days they participate in.
Also designing the experiment is a bore - it requires answering 480 questions to find b and delta.

Further reading

Enhanced by Zemanta


Popular posts from this blog

Moodle <=< Mediawiki SUL integration - first thoughts

downloading folders from google drive.

AWS CloudFormation Pros and Cons