iiinova — where your digital self keeps evolving

Carl Colglazier @carl.cx · 16/09/2025

Subversively using secret Slack commands.

Slackbot
:wave: Hi there, we don’t provide support for this font, so you may run into display issues. You can switch to an officially supported font in Appearance preferences.

000

Carl Colglazier @carl.cx · 22/07/2025

John Snow's Cholera Map

Original map made by John Snow in 1854. Cholera cases are highlighted in black, showing the clusters of cholera cases (indicated by stacked rectangles) in the London epidemic of 1854. The map was created in order to better understand the pattern of cholera spread in the 1854 Broad Street cholera outbreak, which Snow would use as an example of how cholera spread via the fecal-oral route through water systems as opposed to the miasma theory of disease spread. The contaminated pump is located at the intersection of Broad Street and Cambridge Street (now Lexington Street), running into Little Windmill Street. The map marks an important part of the development of epidemiology as a field, and of disease mapping as a whole. [text from Wikipedia]

0120

Carl Colglazier @carl.cx · 22/07/2025

Charles Joseph Minard's map of Napoleon's Russian campaign

Charles Minard's 1869 chart showing the number of men in Napoleon’s 1812 Russian campaign army, their movements, as well as the temperature they encountered on the return path.

190

Carl Colglazier @carl.cx · 14/07/2025

Made a little dashboard in #Quarto and #Shiny to track my job application process. github.com/CarlColglazi...

Dashboard titled “Job Search Dashboard” displays three charts on the left and an application form on the right.

Top-left chart: step line plot with Date (January–May) on the x-axis, Applications (0–100) on the y-axis, cumulative count reaching 100 by mid-May.

Top-center chart: heatmap with weekdays (Mon–Sun) on the y-axis, weekly intervals on the x-axis, color legend “Applications” ranging from 0 (white) to 10 (dark blue).

Bottom-left chart: Sankey diagram with a source node “Applications (100)” flowing to “Screener (57)”, “Interview (27)”, “Offered (11)”, “No Response (22)”, and “Rejected (21)”.

Right panel: text inputs for Company Name, Position, Link; a Status dropdown set to “Applied”; an Application Date field showing “2025-07-14”; and a “Submit Application” button.

130

Carl Colglazier @carl.cx · 06/07/2025

I added a small section on my website homepage that displays a recent post from my Bluesky account. It's powered by a small program I wrote in #golang. github.com/CarlColglazi...

A screenshot of a part of my website with a heading "A Sample (N = 1) from my Bluesky Timeline" displaying a post "Is it time to finally learn vim keybindings?" with a timestamp and a link to view the post.

361

Carl Colglazier @carl.cx · 17/06/2025

GitHub is down and gaslighting me

Screenshot of GitHub.com status page showing operational status for various services. All services are marked "Normal" with green checkmarks except "Copilot," which is marked "Degraded" with a yellow warning icon. Services listed include Git Operations, API Requests, Pull Requests, Packages, Codespaces, Webhooks, Issues, Actions, and Pages.

031

Carl Colglazier @carl.cx · 17/06/2025

This is how your email finds me

Vienna Beef poster for Chicago's Hot Dog, depicting a giant Chicago-style hot dog assembly on Navy Pier with the skyline in the background.

000

Carl Colglazier @carl.cx · 15/06/2025

R package management vs. Python package management

Split paths meme where both options are bad

100

Carl Colglazier @carl.cx · 09/06/2025

#Quarto supports running SQL chunks directly through {DBI}! But does anyone know how to tell #Positron/#VSCode to "run cell" as with the Python/R chunks? #duckdb

Screenshot of a Quarto notebook showing two code chunks. The first is a SQL chunk labeled wiki-counts, which connects to a database using con, selects source_file and the count of rows per source file from the wikis table, groups by source_file, and orders by count in descending order. The second is an R chunk labeled with a timestamp (20250609-093831) that pipes the resulting wiki_count data frame into ggplot, mapping the log of count to the x-axis and displaying a histogram with geom_histogram().

130

Carl Colglazier @carl.cx · 04/06/2025

We can make Mike Tauchman an all-star as a little treat

A side-by-side comparison of American League and National League All-Star player picks by position. The American League includes Guerrero (First Base), Holliday (Second Base), Bregman (Third Base), Witt (Shortstop), Judge, Mullins, and Bellinger (Outfield), Raleigh (Catcher), and Tauchman (Designated Hitter). The National League includes Freeman (First Base), Donovan (Second Base), Muncy (Third Base), Lindor (Shortstop), Wood, Tucker, and Crow-Armstrong (Outfield), Kelly (Catcher), and Ohtani (Designated Hitter). Team logos appear next to each player's name.

000

Carl Colglazier @carl.cx · 04/06/2025

All that's old is new again. YouTube comments on a piece produced by David Cope's (RIP) "Experiments in Musical Intelligence". www.youtube.com/watch?v=CgG1...

If a real person composed this, would you buy it? To me, it is a meaningless patchwork of beautiful phrases. It is an insult to Beethoven, to say nothing of other renowned composers, to suggest that this even has an aesthetic.

My heart and mind are blown. I never imagined I'd see a machine that could compose good music. And the kind of music that I LIKE is a huge bonus. This is so inspirational. I can't even begin to understand how you'd setup a neural net to learn to program music on its own. I'm a computer programmer, but still I am in awe.

I listened to this piece with biased ears. If I was not aware about The origin of this piece I would have appreciated it whole heartedly. My human bias is not willing to accept that a music once associated with maestros only can be created by computer also. How are we going to idolize a computer the same way as Beethoven?

130

Carl Colglazier @carl.cx · 30/05/2025

Don’t speak to me or my little free son ever again

The Glessner House in Chicago with a little free library modeled after it

030

Carl Colglazier @carl.cx · 29/05/2025

Would love to see this website do Richard Prince discourse

A photograph of “Untitled (cowboy)” by Richard Prince — itself a rephotograph — on display at the Art Institute of Chicago

000

Carl Colglazier @carl.cx · 20/05/2025

New Jersey, I didn’t know your game

The Bishop House, an Italian Villa style mansion

A mixed use development at Rutgers University with a statue of a knight in the foreground

000

Carl Colglazier @carl.cx · 16/05/2025

Look at all these funky boys you can write

Image of a keyboard with unicode keys when pressing "option"

Image of a keyboard with unicode keys when pressing "option" and "shift"

100

Carl Colglazier @carl.cx · 13/05/2025

Trying something. Nice how you can just get public atproto data without authenticating. #Obsidian #ObsidianMD

180

Carl Colglazier @carl.cx · 08/05/2025

Came up with this little trick to stop that little white strip from showing up when you scroll in #Quarto #title-block-header:before { content: ""; width: 100vw; height: 100px; margin-top: -100px; background-color: $dark-bg; // your color } (video shows the before)

220

Carl Colglazier @carl.cx · 06/05/2025

Fun to see Qwen3's stream of data if full of sequences like "hmm maybe this, oh wait not that's wrong" Relatable?

A terminal showing the output of a large language model:

Looking up Reveal.js documentation, I see that fragments are controlled via the `data-fragment` attribute, which
specifies the order. The `data-fragment-index` can be used to set the index. However, there's no built-in
attribute to make fragments visible from the start.

So, the solution might involve using JavaScript to trigger the fragment reveal when the slide is shown. For
example, using the `slidechanged` event to check if the current slide has fragments and then reveal them.

Another approach is to use the `fragment` class and CSS to make them visible. But then the fragments would be
visible by default, which might not be desired. However, if the user wants all fragments on a slide to be
visible from the start, that could work.

Wait, but the user might want some fragments to be visible and others not. So maybe using a combination of CSS
and data attributes. For example, adding a class to the fragment that makes it visible, and then using
JavaScript to trigger the reveal.

Alternatively, using the `data-fragment-visible` attribute. Wait, perhaps the user is referring to the
`data-fragment-visible` attribute, which I'm not sure about. Let me think again.

In Reveal.js, when you use the `fragment` class, the elements are hidden by default. To make them visible, you
can use the `data-fragment` attribute with a number, and then the reveal.js will show them in order. But if you
want them to be visible from the start, you can use the `data-fragment-visible` attribute, but I don't think
that's a standard attribute.

Wait, maybe the user is referring to the `data-fragment-step` attribute. No, that's not it.

000

Carl Colglazier @carl.cx · 28/04/2025

"If you're explaining, you're losing" is a provocative and oft-quoted phrase often attributed to Ronald Reagan, but I really can't find any information about its history or origin (example of use in pic).

A tweet from Alexandria Ocasio-Cortez:

A lot of political ‘experts’ like to claim that “if you’re explaining, you’re losing,” but I’ve always thought that was misguided.

Sure - if your explanation is boring & bad then yeah, you’re losing.

But if it’s compelling and good, then the more you educate, the more you win.

020

Carl Colglazier @carl.cx · 17/04/2025

The current openai model

010

Carl Colglazier @carl.cx · 07/04/2025

My final #NCAAWBB rankings for the 2024-2025 season.

Table ranking the top 25 women’s college basketball teams based on Mean Rank, Median Rank, Top Offense percentage, Top Defense percentage, and Win rate. The UConn Huskies rank first with a Mean Rank of 1.1, 94.35% Top Offense, 36.59% Top Defense, and 99.91% Wins. South Carolina Gamecocks follow with a Mean Rank of 2.3 and 99.51% Wins. The table includes teams like Texas, Notre Dame, UCLA, USC, and others, showing variations in offensive and defensive strengths and win percentages. Data shows UConn leading significantly in offensive and defensive metrics.

000

Carl Colglazier @carl.cx · 05/04/2025

Now in the top 5K! Still one very important game to go...

Screenshot of my women's basketball bracket ranking at ESPN, where I am ranked 4,933 with a 99.9 percentile and 1510 points.

010

Carl Colglazier @carl.cx · 01/04/2025

#NCAAWBB model run as we reach the Final Four. UConn are the favorites, but every team has a chance. #MarchMadness

Table with odds of making the finals and winning the championships. UConn is first followed by South Carolina, Texas, and UCLA.

000

Carl Colglazier @carl.cx · 01/04/2025

bsky.app/profile/pfra...

030

Carl Colglazier @carl.cx · 01/04/2025

Went 4/4 on the Final Four with both my men's and women's brackets this year. Women's bracket currently ranked in the top 10K on ESPN! #NCAAWBB #MarchMadness

A screenshot of my standings for my bracket named "Carl's Bracket" on ESPN. It has a rank of 8,835, a PCT of 99.8% and 1190 PTS. I picked UConn as my national champion.

000

Carl Colglazier @carl.cx · 31/03/2025

After many attempts, I got my first win in Twilight Imperium today!

The board game Twilight Imperium with many tiles and pieces on the table

030

Carl Colglazier @carl.cx · 25/03/2025

Updates to my #NCAAWBB model as we reach the Sweet Sixteen. Pulling for the 0.82% outcome!

A table displays projected probabilities for teams in the Women's NCAA Tournament, based on a Monte Carlo simulation. Columns include "Seed," "Team," "E8" (Elite Eight probability), "F4" (Final Four probability), "Final" (Championship game probability), and "Champs" (Championship win probability). The top teams by championship probability are UConn Huskies (37.48%), South Carolina Gamecocks (24.47%), and Texas Longhorns (8.35%). Lower-seeded teams, such as Maryland Terrapins, have much lower probabilities (0.06%). The table is color-coded by team.

120

Carl Colglazier @carl.cx · 21/03/2025

Pre-registering my takes before the women's NCAA basketball tournament begins tomorrow (Wins means the probability of beating an average D1 team). Going with UConn winning my bracket for the second year in a row.

A table with the top ten teams in NCAA women's basketball with columns for team, wins, top offense, and top defense. UConn is number one, followed by South Carolina, Texas, Notre Dame, UCLA, USC, Duke, TCU, West Virginia, and Kansas State.

011

Carl Colglazier @carl.cx · 20/03/2025

TIL you can get a nice view into your data using dplyr::glimpse in #Rstats

Screenshot of console output showing the output of the glimpse function from dplyr where each line lists some data in a column in the table

180

Carl Colglazier @carl.cx · 03/01/2025

This is how I feel when C-SPAN gets to control the cameras

Nicole Kidman looking forward with a projector shining light behind her, from a well-known AMC promo.

011

Carl Colglazier @carl.cx · 18/11/2024

Trying out #Typst output from #QuartoPub for my dissertation. Looks nice! Easier to finagle to fit the university formatting requirements than LaTeX, but not nearly as much package support. Excited to see where the ecosystem will be in a few months.

A statistical table and accompanying text discussing account movement between Mastodon servers. The text highlights that accounts on larger general instances are less likely to remain active compared to those on smaller servers. The table, titled "Moved Accounts," presents coefficients and standard errors for two models (Model A and Model B) with variables like "Sum," "Nonzero," "Smaller server," "Server size (outgoing)," "Open registrations (incoming)," and "Languages match." Significant values are marked with triple asterisks (***) in the "Coef." and "Std.Error" columns. A footnote lists Mastodon server names used in the analysis.

040

Carl Colglazier @carl.cx · 18/11/2024

Touching grass on a drizzly day in Chicago

A foggy urban street with historic and modern buildings in the background. A tall lamppost with an illuminated yellow light stands prominently in the foreground, with cars and a utility truck on the wet street below.

A view of railroad tracks with. The tracks are flanked by beige buildings, and a few orange construction barriers are visible.

A city garden featuring green lawns, flower beds, and pathways bordered by trees. Benches and shrubs are in the foreground, with a large beige building in the background. The scene is framed by overcast skies.

010

Carl Colglazier @carl.cx · 17/11/2024

Yes but from somewhere else

010