Eli Holder
LightNews LightNews
banner
elibryan.bsky.social
Eli Holder
@elibryan.bsky.social
2.6K followers 4.5K following 310 posts
Raleigh. How might we visualize people, fairly and equitably? Data / dataviz / design / psychology / research geek. He / him. Chaotic good. Follows Fizzlethorpe Bristlebane. Design / research / writing at https://3iap.com.
Posts Replies Media Videos
elibryan.bsky.social
Fun sketches → Fun charts! 📊🎉
Pencil sketch of a diverging area chart with annotations Pencil sketch of some kind of jitter plot Pencil sketch of small multiples grid Pencil sketch of growing circle?
September 3, 2025 at 2:35 PM
elibryan.bsky.social
What can we do here?

@elijfinkel.bsky.social speed dating work suggests a much less bleak dating world, vs Tinder's 80/20 dynamic: doi.org/10.70400/NEL...

@jamieamemiya.bsky.social & co suggest counterfactuals to blame systems, not people: doi.org/10.1177/1745...

Which maybe looks like this...
A quote describing Eli J. Finkel’s “Actor, Partner, Relationship” model of attraction. It reads: “Rather than having the 10s match with the 10s... most of us can find a partner who is something close to a 10 for us.” There’s also a figure loosely illustrating this, with lots of people matching with lots of other people. a bar chart showing swiping / liking behavior on tinder between different gender groups text says: “Better: Blame tinder, not women” a play on “Blame systems, not people” The original bar chart showing swiping / liking behavior on tinder between different gender groups, paired with a counterfactual bar chart showing liking behavior in speed dates for different gender groups. The title says “Tinder Distorts Dating Behavior”
August 24, 2025 at 6:29 PM
elibryan.bsky.social
Genetic determinism also shows up in incel belief systems, tangled up with some wild takes on online dating statistics... (e.g. women's suspected urge to "trade up" beyond their "looksmatch").

Preston & co: doi.org/10.1177/1097...
Vallerga & Zurbriggen: doi.org/10.1111/asap...
Slide showing Sydney Sweeney beside the Jamie Miller character from Netflix’s “Adolescence” series Beside Sweeney is text: “Sydney Sweeney is genetically superior” Beside Miller is text “Others are genetically inferior.” This concept of genetic worth is part of both eugenics and the incel belief system Slide showing a fake statistical statement about inequality in dating: “80% of women are only interested in the top 20% of men. So 80% of men have to fight for the bottom 20% of women.” Beside it is a bar chart showing swiping / liking behavior on tinder between different gender groups, which are interpreted among incels as evidence of women’s “hypergamy” A bar chart showing swiping / liking behavior on tinder between different gender groups. A figure loosely illustrating the presumed lookism 80 / 20 rule “80% of women are only interested in the top 20% of men. So 80% of men have to fight for the bottom 20% of women.” To the right are different explanations offered for these disparities (and lack of dates)
August 24, 2025 at 6:29 PM
elibryan.bsky.social
What can we do (for the charts)? Part of the misattributions stem from design choices.

❌ Bar charts increased genetic and personal blame for health disparities.

✅ These "geo-emphasized" jitter plots shifted blame toward systemic explanations.

Links + videos from our last paper: 3iap.com/mbat/
Slide showing a bar chart of public health disparities with the text “Save lives. Stop bar charts.” From my deathly dataviz talk: https://3iap.com/rladies2025/ Slide showing the “geo emphasis” chart from our must be a tuesday study, with text “We can do better” From must be a tuesday study: https://3iap.com/mbat/ How did viewers explain health disparities in our Must Be A Tuesday study (https://3iap.com/mbat/)? It shows that conventional bar charts significantly increased genetic attribution and personal blame. And that the geo-emphasis charts significantly shifted blame from personal toward external attributions. Slide shows a simplisitc chart of outcomes, hiding variability, which says “less of this” And a more complete jitter plot of outcomes, showing variability, which says “more of this”
August 22, 2025 at 4:30 PM
elibryan.bsky.social
Similar stories here:

Good outcomes = good genes
→ "Do Black people really need vaccines or anesthesia?"

Bad outcomes = bad genes
→ "There's nothing we can do to help!"

These charts are obviously well-intended, but ours + other studies find that viewers often misattribute these disparities.
Demonstrating viewers’ explanations for health disparities based on viewing health disparity charts. They often blame people (personal attribution) by explaining the disparities in terms of personal behavior or characteristics “Drinking and smoking”, “too much fried food”, “bad genes.” These may make sense at individual levels, but not for populations. Or you could attribute these disparities to external social determinants of health like “access to care, good food”, “neighborhood pollution”, “structural racism” quotes are paraphrased from qual responses seen in the “Must be a Tuesday” study https://doi.org/10.1109/VIS55277.2024.00021 Tracing a theory of action for how genetic misattributions can build into other beliefs that ultimately lead to tangible real world outcomes, like Black patients being misdiagnosed when presenting with pain. From my deathly dataviz talk: https://3iap.com/rladies2025/ RFK Jr’s statement about Black vs White vaccine schedules. It shows his quote: “We should not be giving Black people the same vaccine schedule that’s given to whites, because their immune system is better than ours.” Two news headlines showing Sen Roger Marshall and Sen Bill Cassidy blaming public health problems on individuals. They read: “GOP Senator Blames Americans For Most Of Their Health Problems” https://www.rollingstone.com/politics/politics-news/roger-marshall-make-america-healthy-again-1235232716/ “Maternal death rate isn’t as bad if you don’t count Black women, GOP senator says” https://www.businessinsider.com/gop-senator-la-outlier-maternal-death-rate-skewed-black-women-2022-5
August 22, 2025 at 4:30 PM
elibryan.bsky.social
Similar story here:

Good outcomes = good behavior (us)
→ Bad outcomes = bad behavior (others)
→ "Should we spend so much $$$ vaccinating people who don't deserve it?"
A quote from RFK Jr “It’s exceedingly difficult for measles to kill a healthy individual ...there’s a connection between those who suffer from measles and individuals who lack proper nutrition or don’t engage in regular physical activity.” A quote from Dr Oz, head of medicaid, “It’s your patriotic duty to be as healthy as you can.” Two quotes from Attwell et al 2018’s focus group study of anti-vaxers. Attwell et al 2018 ‘The Unhealthy Other’ How vaccine rejecting parents construct the vaccinating mainstream. https://doi.org/10.1016/j.vaccine.2018.01.076 News headline announcing cuts to mRNA vaccine research. It reads: RFK Jr cancels $500m in funding for mRNA vaccines for diseases like Covid” https://www.bbc.com/news/articles/c74dzdddvmjo
August 22, 2025 at 4:30 PM
elibryan.bsky.social
How does this work for the lame jeans/genes ad? What's the argument behind the ad? What makes it funny (to some people)? Where does it lead?

Good jeans → Good genes (us) → Bad genes (others) → "haha our genes _are_ great!" → "What if we ranked people based on genetics?!"
Slide with a photo from the American Eagle “good genes” ad with the subtext as actual text: “Sydney Sweeney is genetically superior” Slide on the rhetoric of the “good genes” american eagle ad. It says “Genetic superiority is definitely a real thing. Hint: It isn’t.” It shows the eugenicist Karl Pearson looking excited and pointing to Sydney Sweeney. Slide visualizing the eugenic idea that people can be stack ranked based on some simplistic unilinear concept of genetics. On the left it shows “genetically inferior” with an Elphaba-esque woman. On the right it shows “genetically superior” with the american eagle model Slide with a fake photo of Sydney Sweeney holding up a bottle of bleach spray. It says “Sydney Sweeney supports sterilization... of kitchen surfaces.” The risk of dog-whistle ads is maybe clearer when considering where this kind of rhetoric has led historically, e.g. sterilization programs and genocide.
August 22, 2025 at 4:30 PM
elibryan.bsky.social
What do these have in common? It's not what they "say," it's "what do they suggest?"

They suggest a world where "genes are destiny."

They're blame-framed, so looks, love, and health can seem like immutable personal characteristics, rather than results of big messy social forces.
applying the "woke" critical analysis framework to the jeans ad. What does the ad claim technically? What does it suggest? How might people act on it? Applying the rhetorical analysis framework to the health risk charts we tested in “Must be a Tuesday” https://doi.org/10.1109/VIS55277.2024.00021 It shows a bar chart with heart disease mortality for four different groups (which are stand ins for racial outcomes). It asks what does the chart suggest to some people about why heart disease mortality is higher for people in Group B? applying the framework to a bar chart showing swiping / liking behavior on tinder between different gender groups. What does the chart show technically? What does it suggest? How might people act on it? Applying the framework to RFK Jr’s statements about Black vs White vaccine schedules. It shows his quote: “We should not be giving Black people the same vaccine schedule that’s given to whites, because their immune system is better than ours.” and asks what does this suggest?
August 22, 2025 at 4:30 PM
elibryan.bsky.social
Things to do at SIGGRAPH:

✅ Behind-the-scenes FX secrets for Superman!
✅ Never-before-seen Oscar-worthy animation shorts!!
🤔 90 minutes with me... on "woke dataviz," eugenics, public health charts, fighting data demagoguery?!?

A few new slides from my talk...

📊🧵...
photo from SIGGRAPH 2025 in Vancouver. it shows the word SIGGRAPH in big letters on the expo floor. title slide from my talk: “Woke Dataviz for white guys.” It shows a flirty girl, a bodybuilder, a t rex, motorcycles, rockets, and a dork in military fatigues Slide on the rhetoric of the “good jeans” american eagle ad. It says “Genetic superiority is definitely a real thing. Hint: It isn’t.” It shows the eugenicist Karl Pearson looking excited and pointing to Sydney Sweeney. A collage of well-intended, but accidentally sketchy civic dataviz showing social outcome disparities for health, wealth, education, etc.
August 22, 2025 at 4:30 PM
elibryan.bsky.social
The plots aren't exactly ggdist QDPs. Wanted a similar concrete story (1 dot ~= 1 person), but at 500 dots they're not exactly countable.

Dots are also evenly-spaced on the x-axis, which we needed visually, but @mjskay.com + @racquelf.bsky.social might advise otherwise on precision tradeoffs! 😬
Closeup mockup showing the evenly spaced dots and how dot stacks align between plot rows.
July 30, 2025 at 6:22 PM
elibryan.bsky.social
This is also a (6-month late) response to @datarevelations.com's sagely haiku:

"Make everything gray, except the few things you think should be highlighted"

These WIP mocks show earlier color coding attempts–which I still like stylistically!–but monochrome won on function for guiding attention...
Work in progress mockups showing a version with color coded population segments, instead of the monochrome look that made it into the final report
July 30, 2025 at 6:22 PM
elibryan.bsky.social
TBH this is my first encounter with a "mosaic plot," but the same pairing effect works for showing inequality.

Still compelling as a quick gut read, but it takes some studying to read precisely. e.g. still not sure what's happening with 1st+women?!

en.wikipedia.org/wiki/Mosaic_...
Mosaic plots from wikipedia of expected titanic survival rates where "yes" (light gray). This shows proportionate outcomes where equal portions of all groups survive. Mosaic plots from wikipedia of actual titanic survival rates where "yes" (light gray). This shows disproportionate outcomes where equal portions of all groups survive.
June 25, 2025 at 2:31 PM
elibryan.bsky.social
Student growth charts from @damianbetebenner.bsky.social.

I like how seeing them as a pair (left is baseline, right is after intervention) gives an immediate gist sense of changed proportions.

dbetebenner.github.io/NCSA_2025_Co...
Growth as an Equity Indicator (Baseline Scenario). The mosaic plot depicts percentages of students demonstrating different levels of growth based upon the growth needed by the student. The widths of the columns represent the current percentage of students in each category (in this figure 50%, 25% and 25% have growth goals of > 1 Year’s Growth, 1 Year’s Growth, and < 1 Year’s Growth, respectively. The heights of the rectangles represent the percentage of students in each row category demonstrating that amount of growth (uniform distribution of growth by prior achievement). This shows proportionate results. Same chart as a "good scenario" This figure shows a good scenario with regard to equity. The growth associated with the students needing the equity boost to catch-up shows a preponderance of students demonstration a least a year’s worth of learning. Decreasing the percentage of students demonstrating less than a year’s worth of learning is key.
June 25, 2025 at 2:31 PM
elibryan.bsky.social
PSA: This button on Google Cloud costs $16 per click.

Ask me how I learned...
Screenshot of Google cloud's Vertex AI / "Media Studio" / Veo 2 tool annotated with big red letters saying "$16 per click" and pointing to a submit button. If you follow the page's prompt to "Try this: "Pancakes with maple syrup, cinematic food advertisement b-roll, dynamic."" It will spit out four 8-second clips of oddly rendered pancakes and bill you $16, which is quite a bit more expensive than actual pancakes.
June 20, 2025 at 6:06 PM
elibryan.bsky.social
I didn't want to like it, but recent "vibe coding" models are disturbingly impressive?! And kinda fun for prototyping?! 😬

I'm not new to glsl / three.js / etc, but I was surprised at how much it sped up my learning curve for testing new algorithms / approaches?
to chat gpt: "can you write glsl to show a surface (a rectangle) that's being twisted almost like a dna helix?" Not quite what I wanted, but I didn't realize it could spit out complete, working GLSL? to chatgpt: "let's use glsl to draw 10 lines on the screen, starting from different points on the bottom, but converging toward the top center. Their paths wobble a little bit, but they never overlap." The "non-overlap" solution it gave was surprising and pretty clever.. it said "By giving each line a different phase (here phase = i * 0.7) and choosing AMPLITUDE small compared to the horizontal spacing (1/(NUM_LINES−1)), they never cross." A bit more involved with Windsurf + Gemini, but this was the fastest I've ever gotten three.js up and running. I asked it to generate a foldy flat oval surface and it gave an algorithm for potato chip geometry using ParametricGeometry. It struggled with cell shading, but to be fair, so did I... Then maybe most impressive was how I could give a super vague description of the composition and it generated something reasonably close... with parameters that could be tweaked more to produce something actually reasonable ("okay let's change the layout again and add more of the petals and have them randomly fall down the screen vertically like someone was above the screen dropping them") Asked the robot to turn a noisy black and white image into a flow field then trace the paths without intersecting. This took lots of manual tweaking and a bit of prior knowledge, but still much faster than I could've done on my own. The biggest help was letting me test a "topology" approach that absolutely didn't work, but instead of a day-long rabbit hole, I was able to rule it out in like 20-30 minutes...
May 23, 2025 at 3:20 PM
elibryan.bsky.social
🎆📊 Big win for civic dataviz! NAEP now shows student group scores as full distributions. Even for mass-audience reports, we're not stuck w/ conventional (misleading) bars.

Proud to have a tiny role in this (+ @lesvillegs.bsky.social @rsavoie.bsky.social)

www.nationsreportcard.gov/reports/math...
Postcard from NAEP with icon of a histogram. It says "2024 nation's report card distribution charts. View how NAEP performance varies within specific student groups. NAEP: National Assessment of Educational Progress"
May 5, 2025 at 3:41 PM
elibryan.bsky.social
Had a blast last week at @ncme38.bsky.social #NCME25.

Felt a bit surreal presenting now that "equity" research is forbidden knowledge, but was great to see the measurement community engaging with this and the overwhelmingly positive feedback was reassuring.
talk summary slide. It says "Improving Score Reporting of Student Group Results to Better Support Educational Equity" and lists the panelists. Teaser slide from my talk, asking "Does reporting on achievement gaps motivating closing them?" It shows a screenshots of a news report highlighting racial disparities for a local school district, from Quinn 2020 Experimental Effects of “Achievement Gap” News Reporting on Viewers’ Racial Stereotypes, Inequality Explanations, and Inequality Prioritization Teaser slide from my talk, asking"Do school rankings help families find their way to better schools?" It shows a screenshot of a school ranking website. Teaser slide from my talk on the guarantees made by conventional dataviz guidelines
April 28, 2025 at 7:38 PM
elibryan.bsky.social
I am excited to announce I have been selected to be a judge for this year's Information is Beautiful Awards - a celebration of global excellence in data visualization!

I look forward to reviewing your finest jitter plots.

www.informationisbeautifulawards.com/news/640-mee...
IIB awards judge postcard. It says "I'm a Judge! Information is Beautiful Awards 2024"
February 27, 2025 at 4:34 PM
elibryan.bsky.social
Big crowd here already for #TeslaTakedown in Raleigh
February 25, 2025 at 9:14 PM
elibryan.bsky.social
Move fast and break EVERYTHING.
Screenshot of an article that says "Can Tech Get Nuclear Power to Move at AI Speeds?"
February 14, 2025 at 7:59 PM
elibryan.bsky.social
A "mostly fine" Civilization would be such a relief at this point...
screenshot of Civ VII review. It says "Civilization VII is mostly fine on Steam Deck, which is great news"
February 12, 2025 at 12:57 AM
elibryan.bsky.social
Most frustrating: Pretending that polls reflect Trump's actions completely ignores how these attitudes are actively manufactured and, conveniently, ignores how they're shaped by the media. Even reporting on polling could shape attitudes that might show up in the next round of polling.
quote from Isaac Chotiner's interview with Senator Brian Schatz: "And one of the things that Republicans do well is not allow their pollsters to tell them everything. They invent whole stories and then the polling follows." https://www.newyorker.com/news/q-and-a/do-democrats-have-a-plan-brian-schatz
February 10, 2025 at 9:22 PM
elibryan.bsky.social
Vibes-y Qs also invite misinterpretation.

e.g. "is he keeping his promise?" is 🤷 without also asking "What promises do you think he made? What actions do you think he's taken?" So CBS' headline isn't just shaky, it invites Rorschach hot takes like "See! Trump's doing what he promised TO ME."
Headline from CBS' article saying 'Trump has positive approval amid "energetic" opening weeks; seen as doing what he promised
February 10, 2025 at 9:22 PM
elibryan.bsky.social
His Raleigh office is nearby, so went in the morning with my daughter. Called the number on the door in the lobby and got an answering machine. Expected I guess? Went later that day with a larger group, maybe 100-200 people? Big crowd for a Raleigh Tuesday afternoon www.mobilize.us/indivisible/...
People protesting outside Sen Tillis Raleigh office last Tuesday
February 10, 2025 at 5:48 PM
elibryan.bsky.social
Wow, big crowd for a random Raleigh weekday?!
Crowd outside Sen Tillis and Budd's Raleigh NC offices, protesting Trump / Musk / Vought / Project 2025 takeover.
February 4, 2025 at 7:15 PM