show and tell for hackers in DC.

Round 24: Hose Stable Hattery

14 Sep 2015

Here are the Hackpad presentation notes from Round 24: Hose Stable Hattery.


Speaker #1: Eric Haengel

Homemade Nintendo Wii Remote

Eric built a prototype of a device that contains all the functionality of a Wii Remote. With parts duct-tapped to cardboard, it may not look like much but this little device can measure acceleration, orientation (with a gyroscope), direction (via a magnetic compass) and it can communicate all of that with a wired or BlueTooth connection. Eric (and Travis) want to ultimately use this device to measure and compare martial arts moves – record a master and see how close you can get!

The key component (in addition to the Arduino Uno) was the

FLORA 9-DOF

you can build your own – tutorials are on the AdaFruit site!


Speaker #2: Anna Petrone

Share that Cab!

Not only does waiting for a cab suck, we suck at waiting! Anna took data of DC taxicab wait-times and drop off locations from Union Station and wanted to see if she could optimize things by pooling passengers. Turns out you can, and to great effect! She made a few crude estimations (grouped passengers by the arc of their final direction) and only used two weeks of data, but the preliminary results seem to show that there is a huge inefficiency when we wait for cabs. She created an “inconvenience factor” as a metric for determining if a ride was infeasible to share and grouped passengers to minimize this metric.

When is the best time to share a cab? Friday at 6PM. Now we know.


Speaker #3: Travis Hoppe

Godwin’s Law

Slides

Godwin’s Law: “If an online discussion (regardless of topic or scope) goes on long enough, sooner or later someone will compare someone or something to Hitler or Nazism.” Implicit in the law is that a conversation is effectively over once someone makes a comparison to one of the worst genocides in history.

Too bad no one tested this so-called “law”… until now.

Travis downloaded a dataset containing almost every reddit users comment on Reddit (100GB compressed, over 1 TB of data) and analyzed it for comments referring to Nazis and Hitler to test the validity of the law. Turns out it isn’t true, both the explict and implict statement of the law. Maybe we aren’t as bad as we think? Or maybe the law isn’t precise enough?

See also: The “Greater Internet Fuckwad Theory”.

Ongoing research: + Does a conversation stop post-Godwin? + What is the quality of a conversation post-Godwin? + Can we try to prevent it? Should we? + Is Reddit pathological or representative?


Speaker #4: Haynes Bunn

New Hampshire State House Election Analysis

Hayes works for America Votes and wanted to know how many people fully completed their ballot in New Hampshire. It’s a good question and an important piece of information for those interested in elections. All of the election data was from Secretary of State’s website, meaning anyone can repeat this experiment for themselves.

New Hampshire has a problem though – a weird amalgamation of districts and number of people on each ballot. 400 candidates for 204 seats – this would take forever to do by hand!

So she wrote a hack that used Python and PostgreSQL to do all her work quickly.

Takeaway? In 2014 84% of people completed all of their ballot!


Speaker #5: Thomas Levine

Random Sample of Lines from a File

Presentation source

Thomas needed to sample random lines from a file, a reasonable request. However, a 3 second wait time just to parse a 1GB file was completely unacceptable! So instead of solving the problem exactly, he made an approximation to sample the file randomly. The results in several-orders-of-magnitude speedup! He even took it one step further and applied the technique to large groups of files using clustering.

Caveat: long lines in files (or any uneven distribution) breaks the underlying assumptions of the program! File sampler beware!


Speaker #6: Ben Klemens

Twitter Bot!

Ben is an economist at the Treasury, and as a great favor promised there would be no math in his talk. Instead he presented perhaps the most despondent twitter bot ever created, a tweet per day from the writings of Armenian monk St. Gregory of Narke.

Live: @GrigorNaregatsi

What say you St. Gregory?

Glad he’s got a positive outlook on things!


Thanks to everyone who presented, everyone who attended, @svt827 and @metasemantic for the writeup, @LauraNLorenz for the pictures, and of course thanks to our favorite WeWork for hosting!

Round 25: Percival’s Pernicious Parsnips is already scheduled, so RSVP and sign up to present!