danyelf.bsky.social
Data visualization; user experience with data analysis; general joyful data nerdery
192 posts
870 followers
431 following
Discussion Master
comment in response to
post
That seems like a potentially interesting study then: two different measures of political influence on covid safety (gubernatorial, population) and a clear outcome variable.
comment in response to
post
I think the initial premise was that gubernatorial partisanship leads to policies, like masking rules and school closings, might have been the driving factor. It’s fascinating to realize that population behavior outweighs that.
comment in response to
post
(I certainly agree that covid-consciousness is a thing; I'm just looking around a busy coffee shop with people drifting in for an hour of work at a time without masks, and so I figure it's something else.)
comment in response to
post
I'm not convinced it's COVID-consciousness that's doing it -- I think it's largely that people got better at working from home. Even though there are real advantages to community spaces, it might be hard to quantify the cost of commute & hundreds of dollars per month membership.
comment in response to
post
It might be — MSR isn’t meant to be a product team, and tech transfer can be indifferent.
But the quality of the support model doesn’t reflect on whether the technique has value.
comment in response to
post
Ah, interesting. That makes it easy to compare the wedges to each other — 27 is twice 13; 60 is twice 27 — but makes it more challenging to see how they add. But maybe that’s not an important task, because it’s indifferently supported by a pie or stacked bar.
Stand by. I need to do some doodling.
comment in response to
post
Hm. Vornoi pie charts, hex maps, chorpleths… I feel like there are lots of ways (of differing effectiveness) to slice an area into segments
comment in response to
post
What do you mean by “stacks”?
Pie charts aren’t quite stacks. You can do piles of components , like in Microsoft’s SandDance. You can do blow-up scales to show fractions.
comment in response to
post
It’s a recipe for Tshrab, a traditional teeb pous.
comment in response to
post
Or, alternate phrasing: most people, most of the time, are mostly producing highly predictable output that compresses surprisingly well.
comment in response to
post
I think if we weren’t trying to turn the entire economy into AI, the doubters would be able to be more excited about the bear dancing.
comment in response to
post
The divergence comes from the fact that one side is saying “in two years, we will no longer have dancers so we’re tearing down the Bolshoi, firing the ballet teachers, and investing heavily in bear snacks.”
comment in response to
post
“But shouldn’t my axis start at 0?” Maybe, when zero is a meaningful baseline (“aligned with an axis”); less so when it’s arbitrary (“more blueish”)
comment in response to
post
Worse, the range you are illustrating looks like it goes through only about 30% variance. Since your goal is to show difference between states, you should narrow your color scale to match the data
comment in response to
post
Divergent palettes are meaningful when the pivot point means something: growth/loss, above/below sea level, and so on. There’s no pivot point at 50% knowledge of climate change.
comment in response to
post
Google photos even has a person walking back from jail. My god.
comment in response to
post
Interesting. I tend to prefer to *not* assume culturally-specific things if I can avoid them. Maybe all Finland does in fact hop into a Sauna every afternoon -- but maybe the dip is from a regular commute, or a quirk in how house heating works.
comment in response to
post
Do you know what time those spikes are? Do they look different from other cities or other seasons? Does Helsinki offer a price break on electricity after a certain hour?
comment in response to
post
“We cannot call this a class, lest a person who is not a member of the class is mistaken for one who is”?
comment in response to
post
The non-officer person was struck during an officer-involved bullet-involved incident.
comment in response to
post
Wow. A woman on the internet didn’t love dozens of men “well, actually”Ing to her when she was actually right.
I am shook.
comment in response to
post
But that doesn’t make Leckie’s underlying point wrong. LLMs are giving you a different experience than search engines and they do a different thing and cannot be substituted.
I’m genuinely unsure why this is a hill you seem dedicated to dying on.
comment in response to
post
I feel like “Google put in a possibly-inaccurate summary box, clearly labeled as such” is different from “I ask my LLM a question and if I’m lucky it does a search and replies accurately.”
You might be technically correct that these are the same underlying philosophy.
comment in response to
post
It’s a long and glorious tradition
en.m.wikipedia.org/wiki/Fictiti...
comment in response to
post
As in, creating URLs that don’t (and never did) exist, for reasons other than copyright probing?
searchengineland.com/google-bing-...
comment in response to
post
Google has been using LLMs to deliver search results for twenty years? Now I’m just confused.
What parts of this aren’t true?
comment in response to
post
I think it is at least debatable to describe a thing as a “search engine” that sometimes returns non-results and sometimes summarizes them falsely. In the same way a librarians might not call a set of books an “encyclopedia” if 20% of its entires are false
comment in response to
post
I think we’re developing a (novel!) taxonomy of search engines. A tool that spiders the web but only ever returns “zombo.com.” One that then summarizes its results, often incorrectly. One that doesn’t spider, but acts as if it did.
In 2015, none of those were search engines.
comment in response to
post
Thanks for the extra context!
comment in response to
post
I’m curious why it was ICE doing that. Shouldn’t a federal domestic warrant against a citizen be FBI?
comment in response to
post
Clearly it’s (W)eed, (W)acky, (J)esus (D)eath
comment in response to
post
Without disagreeing with your broader point … Some number of campers will discover the joy of the outdoors; many of them will be grateful to have a bowline at their fingertips; some smaller number of them will someday be caught in a storm and grateful to make a shelter.
comment in response to
post
Your home page may feel transient — university.edu goes away; geocities ages out; you may not want your professional page to point to your weird art project or vice versa. I suspect linktree feels like neutral ground to some.
comment in response to
post
Future studies will figure out cost and value. Economists will argue. Public health recommendations will be updated. Cities will update their water processes. This is good and normal. This is how I want science to go.
comment in response to
post
I’m not shutting anything down. I’m mostly just confused. You’re showing me all the signs of perfectly normal science being done: the research on possible thyroid implications of fluoridation will be cited in future public health examinations of fluoridation effects.
comment in response to
post
There are many things we don't know 100%, and science has many active debates on them. People are still publishing reviews of flouridation effects -- even in Nature as of 2024. But you started off not by saying "we don't know 100%" but "the science ... is not good". What makes you say so?
comment in response to
post
This is a genuine question: I feel like this is the standard phrasing which the “reasonable” anti-maskers and anti-vaxxers use—“we need more science, public health oversimplifies, we need more communication, maybe just less of the thing would be better.”
How do you separate yourself?
comment in response to
post
In the 1980s, there were a lot of really good natural experiments on this because fluoridation was introduced at different times at different places, while public health campaigns for brushing were more universal.
Do you have a non-Ripperian reason for suspecting that all the science is wrong?
comment in response to
post
In practice, those people spend a lot of time sitting on their hands (or being assholes) when someone else in the department does work that presumes evolution or gives a lecture or teaches a class.
comment in response to
post
This is why it does no good to cede human rights ground: the right periodically makes millions of people suddenly decide that previously mostly uncontroversial topics like evolution, abortion, vaccines, or women’s sports, are suddenly the Most Important Thing Ever.
comment in response to
post
Really? Everything I’ve seen has looked something like this study (from 1989) showing 30% or so improvements.
pubmed.ncbi.nlm.nih.gov/2681730/
comment in response to
post
The right communications machine has shown it is capable of galvanizing supposed moderates over left bogeymen since the civil rights era. That doesn’t mean the bogeyman in question has power.
(After all, there were no actual furry kids using litter boxes in classrooms; that panic was made up.)
comment in response to
post
That’s interesting—I hadn’t realized there were groups that switched allegiances so thoroughly.
comment in response to
post
Sure. The far right has the secretary of health and human services. During Trump 1, he suggested alternative therapies to COVID from the White House desk.
Can you name a single policy that was altered by a crunchy essential oil lefty, or a person of political prominence so advocating?
comment in response to
post
Has the far left crunchy essential oil crowd ever had any meaningful political power or influence? Or is it merely an interesting observation that this population exists?
comment in response to
post
This study would have coded that as fact based , I suspect, because it used lots of (false) factual words
comment in response to
post
I’m so proud! Other countries are now issuing travel advisories!
www.fodors.com/news/news/li...
comment in response to
post
There are a lot of people who couldn't shift their plans quickly enough. I suspect this number will continue to drop in the coming months.
comment in response to
post
Wait, you mean they successfully built a reproducible model for running their code? What did they have before?
At any other startup, "we managed to get the code running a second time" would probably not get a headline.
comment in response to
post
Reminds me of the COVID Times covers a bit.