Why Stacey Abrams Shouldn’t Concede, by the numbers

Journalist have already covered the massive voter suppression orchestrated by Brian Kemp.  It’s some next level shady stuff.  I had another piece ready to go about all the things the Abrams team could look for to show fraud, but as it turns out, basic math is all they need.  Georgia law dictates a mandatory runoff if no candidate exceeds 50%.

Given it’s not random why some votes haven’t been counted and given Kemp is very much opposed to having them counted, I think we can assume they are balanced in Abrams favour.  So how much does it need to be balanced in her favour for to spark a runoff? That answer depends how many ballots are outstanding. Here are a few scenarios I quickly whipped up using this very simple calculator.  This calculator actually slightly favours Kemp in that you only put in the % support Abrams would need to have and it assumes Kemp takes all the remaining support, when in reality, Metz could take some of those votes.  This was to simplify the calculation and since I’m a known liberal conspirator, I’m going to give Kemp this slight edge in the calculation.

As of this writing, Kemp has just shy of 50.3%.

There is a dispute between the number of provisional ballots in existence:

  1. Kemp says there are only 20k provisional ballots
  2. Abrams team says they collected  31k names from counties of provisional ballots casts.
  3. A judge in Dougherty county has ruled GA accepts ballots postmarked by election day
  4. Still unclear how many more outstanding ballots there are

Basically as long as Abrams has roughly 60% support on enough of the outstanding ballots, there will be an automatic runoff. I built a quick interactive version of my calculator which I published to Tableau Public.  You can use it to create your own scenarios and see what the outcome would be.  

Here are some probable scenarios that would spark and automatic runoff:

  • There are at least 100k outstanding ballots to which Abrams has at least 65% support
    • If there are at least 150k outstanding ballots, she only needs 57.5% support
  • Even if they only count the roughly 31k ballots team Abrams has found collecting county records, she only needs to exceed 85% support amongst that small group to force an automatic runoff

Real journalists have covered the multitude of reasons why this campaign would be declared rigged in any other part of the world so I don’t need to cover that.  What I can tell you is I volunteered to transcribe those photos of county records this weekend to help team Abrams assemble the count. There is nothing random about the names on those lists.  There is a reason Brian Kemp doesn’t want them counted. Don’t let him steal this election in plain sight.

Math, Methodology, Meritocracy and Other Misused M Words

Today Fortune magazine launched their annual Business Person of the Year ranking list.  The list contained 5 women out of 50.  Fortune’s digital editor @aarontask was quick to respond to the inevitable feminist outrage with the following tweet:

“What % would be acceptable to you? Again, we have more than 2x the % of female execs in the S&P 500.”

Mr. Task then did what I have seen just one too many times to stay quiet.  He hid behind the M word.  In this case it was “Methodology” but other favourites in the community include “Math” and “Meritocracy”.


“Have you looked at the methodology? 10 metrics: Financial results = backbone of analysis.

This is typical and not surprising or new but this time it infuriated me enough to speak out.

Let’s start with the basic premise of the word “Methodology”- Websters defines Methodology as “a set of methods, rules, or ideas that are important in a science or art : a particular procedure or set of procedures”

The definition says “science” in it, wow Fortune must be using all kinds of well defined metrics to back up this scientific claim.  As a professional math nerd, I looked forward to hearing a mathematical explanation as to how us women, with our tiny little girl brains should understand that 10% representation is actually quite generous.  This took my back to my days as a Math T.A. while completing my Masters of Science in Mathematics.  I learned a thing or two about grading methodologies back then.  Like how if you made me laugh while reading a foot high stack of math exams, I would throw you an extra half point for your effort.

I also learned to be very very wary of any “methodologies” that were missing sufficient criteria to solve the problem.  Usually that implies either cheating or guessing.  Much like how elementary school students get zero for answers that don’t show their work, I rate Fortune’s “Methodology” an F minus.  The minus comes from having the audacity to talk down to us by pretending there is anything scientific about what defines your process.

Today, Sheryl Sandberg was quoted on CNN saying:

“There are two options.

1. That men are far superior to women and deserve 95% of top level positions.
2. That there is systematic oppression.

Pick One.”

Let’s take a step back from the systematic oppression many thought leaders believe are endemic in our society and stick to the “tangible facts” that Fortune is so fond of:

Tangible Fact #1:  The formula given by Fortune lacks sufficient data to be meaningful.  They listed a bunch of metrics such as 12 and 36 month increases in revenue and profit but failed to place a coefficient or weight in front of them.  Meaning on this scientific scale they failed to say just how much each of the listed performance metrics contribute to the overall ranking.  The icing on the cake is the methodology finishes with a list of what they are calling “non-financial elements” (read:  subjective gut feel) such as influence, leadership style and strategic initiatives which “played a part in our evaluation as well” in the rankings.

I don’t expect Mr. Task to be able to compete with mathematical intellect, so let me mansplain back to him with a parallel example my 9 year old nieces would call BS on.  Mr Task has baked a cake by adding unknown quantities of flour, eggs, baking powder and vanilla and a dash of each of his favourite secret ingredients.  Therefore I have scientifically proven Mr. Task’s cake recipe is the best recipe in the entire country.

Mmm tastes delicious and not condescending at all.

Tangible Fact #2:  The Fortune criteria shows a clear lack of organization ambidexterity.  The concept of Exploitation vs. Exploration is one of the most fundamental challenges facing businesses with the emergence of a data culture.  The best predictor of future success is past success.  This is typically true assuming your definition of future success is the same as past success and provided you are out of business before the diminishing returns of this approach start to backfire.  To avoid these diminishing returns one needs to consider the value of the future.

Consider the popular Facebook People You May Know feature.  This is a classic example of how Facebook is incorporating exploration into their daily practices.  Facebook could sell that prime advertising space to the right of your feed (and often does) but they also sometimes choose to forfeit income today on a bet that increasing their social web connections will make their user base happier and more valuable in the future.  Believe me when I tell you Facebook is not doing this because it makes them feel warm and fuzzy to connect people, it’s because they are betting big on growth.

This example is even more clear if we restrict current and future revenue to the same criteria.  I built and ran the Data Science team at Kobo, a global eReading company. While the recommender systems didn’t fall under my purview, I had the opportunity to discuss this problem at length with many smart researchers.  Kobo sells eBooks.  The more they sell, the more revenue they make.  Books bought through the Kobo recommendation engine compose a non-trivial portion of overall sales.  At the onset, the highest recommendation conversion rates will no doubt be for books similar to ones the customer has already read.  That’s great until you run out of similar books or your customer is finally bored of reading trashy romance novels.  For that reason, Kobo, Amazon, Netflix and any other companies with recommendation systems worth a damn include some component of exploration to search for different items their customers may be interested in.  This also helps control the vicious feedback loops that permeate these types of problems.

In this case, Fortune is playing an active role in perpetuating the feedback loop as all 50 leaders on this list will benefit from press, increase their name recognition and likely bring back even higher revenue in the future as a result.  Since only 10% of the leaders on this list are women, a disproportionate number of men will benefit from the exposure and therefore have a leg up in future “best of” lists and at the top of more executive job searches thus making them even more likely to get jobs with companies at the top of the S&P 500.

Fortune is showcasing themselves to be a dying dinosaur by failing to explore other measures of success.  There is good news Fortune, for a very reasonable price you can hire a woman like me to help your “methodology” graduate to something less embarrassingly rudimentary.

Data porn for the inner nerd in all of us

Yesterday I stumbled across this Washington Post blog using data and maps to explain the world.

After going through all 40 maps I have concluded that my home and native land of Canada is the best.  (Obviously).

From a data perspective, not all maps are equally interesting.  The most interesting pieces, tend to have the most subjective insights such as the best and worse places to be born, where the detailed explanation still leaves a lot of questions to be answered.

All and all an interesting light read though I would have loved if the post took it one step further and made the maps interactive over time.  At least a few of the charts have the data to support that and it would be very interesting to visually watch history evolve on a map.