Jacob Hall

howdy! this is my weblog.
please leave your shoes by the door

📧    📷    👨‍💻    🛋️    🐘

about - linkroll - ideas

👣 Barefoot FAQs

I've been going barefoot for many years now, since my ~sophomore year of high school. I don't think I've ever met anyone who's gone barefoot as much as I have, so as you might imagine I get a lot of questions about it. Writing this post is, as usual, mostly for my own benefit. However, I'll try to make it interesting to anyone who'd like to read some stories about my experiences!

Why do you go barefoot?

In short, I find it more comfortable.

Sometimes people will ask me if I go barefoot to be "grounded" or for a heightened sense of spirituality. I honestly don't know much about grounding, and it's not something I ever think about. As an atheist, I don't find much religious or spiritual meaning in going barefoot, but it's cool that other people do!

People have asserted that I go barefoot to attract more attention. I love meeting and talking to people, and freely admit to expressing myself in ways that spur conversations! I honestly forget that It's Weird To Go Barefoot. Along with the way I dress, do my hair..., the way I express myself is genuine. It's ok if you don't think my self-expression is attractive or sexy or tasteful, but I do!

When do you go barefoot?

In the warm season, which for me in Williamsburg, Virginia is ~March-October. I plan to move somewhere warmer after I graduate from college, so that I can go barefoot all year and generally avoid cold weather.

Where do you go barefoot?

Wherever I'm allowed to, which is most everywhere. There are no laws (to my knowledge) that disallow it, so I'm only banned from places that intentionally ban me. I've written some stories below about time I've gotten asked to leave different places.

Have you ever been kicked out of somewhere?

Yes, plenty of times.

During the 2018 March for Our Lives protest, a large group of students from my secondary school marched from our school to the Capitol Building. Once things had wrapped up, I really had to pee. We were close to the National Museum of the American Indian, and knowing that they have a public bathroom I popped in for a minute. The security stopped me, and asked if I had shoes. When I said that I did not and just needed to use the bathroom, their supervisor said that I may not enter the museum without shoes on. I'd peed myself a little by the time I got to the Botanic Garden bathroom!

When I was a student at James Madison University I was frequently asked to leave the dining halls, but I don't recall any issues going barefoot to class, except for a bio lab professor who required closed-toe shoes in class. The following year I transferred to William & Mary, and started having similar problems with dining hall employees asking me to leave. I filed a complaint with the Office of Compliance & Equity, who forwarded my case to a Human Resources Manager for William & Mary Dining Services. He kindly investigated the issue for me, and concluded that there are no laws, "Food Safety issues," or campus policies that require shoes. Since then I've been asked to leave dining halls a number of times, but I've either had my phone handy to show them that email, or they've been intercepted by a manager.

When an employee at a store or restaurant asks me to leave, I try not to make too much of a fuss. Maybe it could be seen as discrimination, but I want to respect the wishes of business owners. It is their property after all! It doesn't happen very often though, and for every time it does, there's another time where an employee apologizes to me about a spill on aisle X or something like that. Once the owner of a Kebab restaurant came out from the back and profusely apologized to me because she hadn't cleaned the floor that day! Writing that story years later brings a smile to my face 😁

Have you ever hurt your feet?

Never very badly. My feet are pretty tough at this point, and I watch the ground as I walk! I can remember two minor cuts I've gotten on my feet that stopped bleeding very quickly. I can say with certainty that I've cut my hands way more times! I'm very confident at this point that I know the risk I'm taking: I've stepped on a lot of things and understand how they can feel or hurt me. Even if I do injure my feet severely, I doubt I'll stop going barefoot as long as I'm still able.

⌨️ Simple HTML code highlighting workflow with vim

Lately I've wanted to post code snippets to my website, but I want them to look pretty and highlighted instead of just using boring old <pre> tags. I know a lot of people use fancy scripts to highlight text client-side, but one of the rules I've set for myself in building this website is to avoid Javascript. Besides, is all that effort and computation really needed for some colored text? So I decided to find a workflow in which I can export my code into pre-formatted HTML.

I recently put some effort into customizing vim, and after trying a bunch of themes settled on Edge. Staring at my beautifully-highlighted code in vim, I wondered if I could just replicate the same effect on my website. I quickly found the vim TOhtml function, which does exactly that! All you have to do is run TOhtml in vim, and it outputs a formatted HTML file complete with CSS rules that color different types of syntax exactly how it's displayed in the editor. Hoorah!

After messing with TOhtml for a while, I realized that it was not actually outputting the same syntax highlighting that I was seeing in my terminal. For example, functions appeared light blue to me in vim, but the exported HTML did not discern them at all. It took me a long time to realize that the issue was with my color scheme, Edge! I have no idea why, but TOhtml does not play well with Edge and ignores some of its syntax highlighting rules. It also seems to forget that I have true colors turned on! I tried to run TOhtml using the popular Solarized scheme, and it worked without hitch!

...but I want to use the Edge color scheme. And besides, if I ever want to switch color schemes in the future, I don't want to worry about their compatibility with my current syntax highlighting workflow. In my .vimrc I've created a variable that defines my current color scheme of choice:

" Use edge color scheme: https://github.com/sainnhe/edge
let g:usual_colorscheme = "edge"

" Set our colorscheme to the one I've defined above
execute "colorscheme ".g:usual_colorscheme

By doing that, I can declare a function that temporarily sets my color scheme to solarized when I want to use TOhtml, and then define a command to run it:

" Function to output better syntax-highlighted HTML using solarized
function SolarizedTOhtml()
	colorscheme solarized
	execute "colorscheme ".g:usual_colorscheme

command NEWhtml call SolarizedTOhtml()

It was humorously difficult to define suitable CSS rules for these vimscript snippets, but I think they look very nice considering how simple this system is. Since I'm only exporting code using Solarized now, the same CSS rules should work on future code snippets without much modification, and I'll be able to adjust them all uniformly with ease. The only other adjustments I've made to this setup are the following two settings in my .vimrc:

" For :TOhtml output, remove line numbers
 let g:html_number_lines = 0

" For :TOhtml output, ignore code folding
 let g:html_ignore_folding = 1

Have a beautiful day!

📍 Mapping Education Development Projects in Afghanistan

This is the first time I've used R and LaTeX together, using a package called knitr. I believe papers that include statistical analyses should be fully reproducable, and part of my objective for this project is to learn how to accomplish that. I was partly inspired by this article on Chinese investments (also using AidData data), available as an R notebook 🔓 here. I also used this guide to help navigate the weird world of the maps R package. As I got further into the weeds of plotting the population density data, this Data Carpentry lesson was indispensable.


I requested a bunch of data about Afghanistan's education and forestry development from AidData. The download link for my query is 🔓 here. I decided to focus on the education data for this project, and experiment with overlaying the education project locations over a population density map of Afghanistan. I hope for this map to effectively convey how well-distributed these projects have been relative to populous areas. Have a majority of these projects been in cities, or in a specific region?

Directly converting the population density raster to a data frame ate my laptop's memory and caused R to crash, simply outputting "Killed" in my terminal. I ended up asking for help, and it was suggested that I aggregate the data before conversion. I'm annoyed that R can't handle this data file with better memory management (and I dream of pixel-perfect output!), but the aggregated data still looks very nice.

# Import AidData dataset
# Make sure you edit this path to reflect the file's location on your machine
inscription_data = read.csv("results.csv", header = TRUE)
education_data <- filter(inscription_data,
                         grepl("Afghanistan", recipients),
                         grepl("education", ad_sector_names))

# Import population density data for Afghanistan (2020)
afg_pop <- raster("afg_ppp_2020.tif")
# Theoretically not necessary if you have a lot of RAM overhead 🙄
afg_pop2 <- aggregate(afg_pop, 10)
# This is the line that caused my laptop to crash without the aggregated input
afg_pop_df <- as.data.frame(afg_pop2, xy = TRUE)

afg <- map_data("worldHires", "Afghanistan")

# Let's define a map that plots the population data all pretty
pop_map <- ggplot() +
  # Remove axis labels because it is clear they are lat/long
  theme(axis.title.x = element_blank(),
        axis.title.y = element_blank()) +
  # Add population data
  geom_raster(data = afg_pop_df,
              aes(x = x, y = y, fill = afg_ppp_2020)) +
  # Define scaling for population data
  scale_fill_continuous(name = "Population\nDensity",
              # Poke holes wherever there isn't data
              na.value = NA,
              # Use pretty, colorblind-friendly colours
              type = "viridis",
              # Use logarithmic scale
              trans = "log10",
              # Custom breaks to make key more readable
              breaks = c(0.1,1,5,20,100)) +
  # Add outline of Afghanistan on top for a cleaner edge
  geom_polygon(data = afg,
               aes(x = long, y = lat, group = group),
               fill = NA,
               color = "black")

# Now let's add the education project data to the map, and plot it!
pop_map +
  # Add education project sites
	geom_point(data = education_data,
             aes(x = longitude, y = latitude),
             # Circular points
             shape = 21,
	           size = 1,
             color = "white",
             # Do not fill circles
             fill = NA) +
Population heatmap of Afghanistan with projects overlayed as circles Full-size image

That looks great! I know it's wrong to not have units for the population density scale, and that's something I'd like to fix. The value of each pixel in the original image represented the number of people in the area that it covered, so the numbers on this scale should represent the number of people who live within each of the aggregated chunks. I think this issue would be best solved by calculating the distance each chunk covers in mi/km, and then adjusting the values in the legend to represent people per square mi/km.

That cluster of projects (and people!) in the Northeast is Kabul. How many projects are occuring in the greater Kabul area?

# Coordinates from Wikipedia article on Kabul
kabul_lat <- 34.525278
kabul_long <- 69.178333
radius <- 0.3

# New data frame from education_data with just the rows within radius of those coordinates
near_kabul <- filter(education_data,
                     between(latitude, (kabul_lat - radius), (kabul_lat + radius)),
                     between(longitude, (kabul_long - radius), (kabul_long + radius)))
# The number of projects within that radius can be determined using the following command
# I've suppressed the output of this code chunk so that I can call this command inline
The same population heatmap, with one larger circle around Kabul Full-size image

I think that looks a lot nicer, although it might benefit from more detailed info in the legend about what different sizes of circles indicate. I don't notice anything groundbreaking on this map. There might be fewer projects on the Western side of the country, but I guess that has more to do with the population around Kabul than any other bias. Kabul is nearly 10 times as populous as the second-biggest city, Kandahar, so it is no surprise to see a majority of projects in that region.

Is this really all of the projects? Our education_data variable has `r nrow(education_data)` rows, and even with the `r nrow(near_kabul)` removed around Kabul, are all of them being displayed? I noticed that the dataset has rounded numbers for a lot of the latitude and longitude values, which could mean that many projects in the same town are drawn on top of each other. For example, there are three projects in Badakhshan that are all listed as having the exact same coordinates, despite differing descriptions, years, and dollars spent. There are two methods I'd like to use to mitigate this problem. The first is to cluster points that are within a radius of each other, and draw circles of proportional size to represent each cluster. The second is to calculate the average population density within a radius of each project, and then create a bar chart with each bar representing a range of population densities, and its length representing the number of projects that fall within an area of that density. It might be useful to create a few charts using different radii of average density.

Other variables could be controlled to better represent the differences in aid allocation across Afghanistan. For example, the size of each point on the map could be scaled based on the monetary value of the project it represents, giving the viewer a better idea of where the majority of funds are allocated geographically. Projects in urban areas might receive more funding to serve more people, or maybe the cost of building a school in a rural area would be significantly more expensive, for example. The source of funding, years active, and preexisting school enrollment are all potentially insightful variables to incorporate in a visualization like this.

Data Sources

Goodman, S., BenYishay, A., Lv, Z., & Runfola, D. (2019). GeoQuery: Integrating HPC systems andpublic web-based geospatial data tools. Computers & Geosciences, 122, 103-112. The data from AidData used in this project were downloaded from 🔓 this page.

WorldPop (www.worldpop.org - School of Geography and Environmental Science, University of Southampton; Department of Geography and Geosciences, University of Louisville; Departement de Geographie, Universite de Namur) and Center for International Earth Science Information Network (CIESIN), Columbia University (2018). Global High Resolution Population Denominators Project - Funded by The Bill and Melinda Gates Foundation (OPP1134076). https://dx.doi.org/10.5258/SOTON/WP00645 The data from WorldPop used in this project were downloaded from this page.