I recently took a class at work about using “precise language”. One of my takeaways was to be deliberate about what I mean by each word when I talk about people, particularly their demographics like race or gender. For example, what does it mean to say someone is “diverse”? Often that’s a term used to “other” people and reinforces they they are not the default (which often in the tech industry and world means white male).

When my book club talked about the desire to pick a “diverse author”, I wanted to figure out what we actually mean by that. I.e. what characteristsics do we over-represent with the authors we’ve picked?

The following graph shows each author of a book we’ve read, along with their age, gender, and race (according to quick Google searches for each).

(there are many other characteristics of authors or their books we could consider as part of a desire for “diversity”)

With the exception of the string of 7 white men to start with, we’ve been almost perfectly split male/female authors. But we’re overwhelmingly white and, surprisingly, gen X. Therefore, I believe my desire for a “diveristy” in our authors could be representative of a lack of non-white and younger authors.

Jess Kid is our first millennial and picked before I wrote this, so we took a step in that direction without acknowledging that was a direction we wanted to go. Perhaps this study will help inform us for the next book we pick.

Los Angeles and CA have a housing shortage. Since there are fewer units in supply than demand, housing prices have outpaced incomes. As residents are pushed to cheaper housing further from job centers, commutes get longer and traffic gets worse. California has a large and rapidly growing homeless population and Californians are much more likely to share a room with another adult.

There has been pushback against increasing the allowed density of residential units. A major fear has been that increasing supply means increasing density and changing the character of single-family neighborhoods. “Manhattanization” is often used to describe efforts to increase density.

The analysis below shows that Los Angeles can eliminate its housing shortage without significantly changing the vast majority of residential neighborhoods.

There are 1,790k single family homes in LA County, and 24k multifamily buildings containing 1,560k units, for a total of 3,350k units [1]. The LA housing shortage is described as about 517k units.

Fourplexes can blend in with single-family homes due to their small size and general conformance with existing single-family zoning laws (this is the basis for SB 902, which aims to allow 2-4plexes across the state). A quick image search shows fourplexes look like large single family homes. If 173k single family homes were converted to fourplexes, 519k units would be added, eliminating the housing shortage. That’s a little under 1 in 10 houses, or about one fourplex per block. This isn’t a significant change to any neighborhood.

Graph showing the number of buildings in LA county before and after building 519k units of quadplex.

But that is assuming an even distribution of supply. If we want to improve traffic, housing needs to be near jobs to lower the total people-miles of commuting. The Westside, Beverly Hills, and El Segundo are major job centers and together have about 113k single family homes [2]. We have to look denser than single-story buildings to fit the units in one region of LA:

• The highest density zone (R5) allows one unit per 200 sqft of lot size.
• The average single family home in CA is 0.15 acre, or 6500 sqft.
• In a vastly simplified calculation, each unit of single-family home could be converted to 33 units of R5 housing.
• It would take 16k home->tower conversions to address the shortage.

We could solve the LA housing shortage building 2400 acres of residential towers, which is about the area west of Lincoln in Santa Monica + Venice. Putting high density towers near job centers could significantly change the surrounding area, the small number of changes (<1% of R1 buildings) means a vast minority of LA neighborhoods would be affected. Especially if it’s spread out across the 3 million acres of LA County.

Graph showing the number of buildings in LA county before and after converting 16k R1 homes to R5 housing.

Graph showing the number of buildings in the “Westside” before and after converting 16k R1 homes to R5 housing.

Of course, these are both extreme cases. Skyscrapers are never going to appear in Venice overnight (or possibly ever), nor will a hundred thousand homes get torn down for more density. Medium-density is the “missing middle” that enables walkable neighborhoods and would require much less upzoning than the first case, with a lower impact than the second.

Additionally, this only considers upzoning single-family properties. There are 215k 2-9 family buildings that could be upzoned to medium-density in areas that already have appetites for multifamily housing [1]. Many of the ongoing major housing projects were on non-residential land, like former industrial mills or car dealerships.

It’ll take a combination of changes to address the housing shortage in LA. But none of them will cause LA to look anything like Manhattan.

[1] 2018 LA County Asessor Parcels Data

select units, count(1) c
from Assessor_Parcels_Data_2018
where generalusetype = 'Residential'
group by units
order by units asc;

[2] 2018 LA County Asessor Parcels Data. Everything southwest of 34.079536, -118.404716 is considered the “westside”. (There are better definitions, but this is a simple query).

select units, count(1) c
from Assessor_Parcels_Data_2018
where generalusetype = 'Residential'
and center_lat <= 34.079536
and center_lon <= -118.404716
group by units
order by units asc;

I just finished listening to the first major campaign/arc of The Adventure Zone, a podcast where three brothers and their dad play Dungeons and Dragons, broken up inter a series of stories. It’s very good! Each story definitely felt pretty different, especially the last stories (especially especially The Stolen Century) seemed to have a lot of exposition by the DM and little action by the players. So I set out to see if there were any trends for how much players spoke each stories/episode.

Spoiler alert: Results were in the header image. It shows the fraction that each player spoke in each episode, as defined by numbers of words spoken. The darker shades represend the human players speaking, while the lighter shades represent their characters (or the NPCs for Griffin, the DM).

There doesn’t seem to be a trend across the whole campaign, though I was surprised by how much it varied. The Stolen Century story, which I expected to be very Griffin-heavy, was a pretty average 50% of Griffin + 10% as NPCs. The finale was definitely mostly DM, with Griffin making up 78.7% of words in the penultimate episode.

I was also surprised by how much the fraction of in- and out-of-character changed by episode, even within a story. For example, Petals to the Metal had episode 21 with one of the highest fraction of in-character words (59.3%), then just two episodes later had some of the lowest (4.3%, 5.6%, 9.1%). These correspond with episodes of low-action – episode 21 was basically where the PCs talk to the quest-givers about what’s going on – and the action-heavy battle wagon race.

There’s a lot to look at in the data. Take a look for yourself at the link below.

The data!

Which hugely relies on the incredible effots of the TAZscripts team. (thank you!!)

TAZ Transcribed

And how the script for pulling the data.

github