Jim Cowie
LightNews LightNews
jimcowie.bsky.social
Jim Cowie
@jimcowie.bsky.social
62 followers 110 following 15 posts
Fellow, Berkman Klein Center for Internet & Society at Harvard University (2024-2025, 2025-2026).

Internet History Initiative, DeepMacro, Renesys, Internet measurement, data science, finance, geography, chaos. He/him.
Posts Replies Media Videos
Reposted by Jim Cowie
beidelson.bsky.social
A nice write up about Case Viewer, my legal browser app: hls.harvard.edu/today/harvar...

If you have an iPhone, iPad, or Mac and haven't tried it, you should! Learn more at www.caseviewer.app, follow @caseviewerapp.bsky.social, or download for free from app store: apps.apple.com/app/apple-st...
Harvard Law professor launches Case Viewer app for easier access to court decisions - Harvard Law School
Harvard Law Professor Ben Eidelson tapped into a childhood passion to build a tool for lawyers and law students.
hls.harvard.edu
November 11, 2025 at 1:13 PM
Reposted by Jim Cowie
harvardlil.bsky.social
Podcast: Jack Cushman joins “Pioneers & Pathfinders” to discuss libraries shaping legal tech, digital preservation, and realities of legal AI. Listen: www.seyfarth.com/news-insight...
Pioneers and Pathfinders: Jack Cushman
Today, we’re joined by Jack Cushman, director of the Harvard Library Innovation Lab, where he and his team are reimagining how library principles can shape the future of legal technology. Jack is a…
www.seyfarth.com
October 31, 2025 at 1:40 PM
Reposted by Jim Cowie
gretchenkrueger.bsky.social
I’m thrilled to share that I’ve been helping out my brother David who is starting a new org, Evitable.com, focused on informing and organizing the public around societal-scale risks and harms of AI, and countering industry narratives of AI inevitability and acceleration! 1/n
Evitable
Evitable.com
October 29, 2025 at 5:31 PM
Reposted by Jim Cowie
meredithmeredith.bsky.social
The question isn’t "why does Signal use AWS?" It’s to look at the infrastructural requirements of any global, real-time, mass comms platform and ask how it is that we got to a place where there’s no realistic alternative to AWS and the other hyperscalers. 3/
October 27, 2025 at 10:38 AM
Reposted by Jim Cowie
tedunderwood.com
If populism is (as this argues) a rebellion against System 2 thinking—which requires symbols and remains frustratingly hard for most of us—it seems we’re stuck with it forever.

If only we could make some kind of, idk, cognitive prosthesis for symbolic reasoning? Yeah, never mind. Sounds hard. +
Populism fast and slow
It is natural that a person who is both concerned by the rise of right-wing populism and possessed of a bookish disposition might turn to the academic political science literature in search of a bette...
josephheath.substack.com
October 26, 2025 at 1:31 PM
Reposted by Jim Cowie
juliaangwin.com
Come work with me! I'm hiring a data scientist to join me at @harvardkennedy.bsky.social @shorensteinctr.bsky.social

careers.harvard.edu/job/senior-d...
October 25, 2025 at 1:24 PM
Reposted by Jim Cowie
eldomador.bsky.social
Amazon has blamed today's massive outage on the "DNS resolution of the DynamoDB API endpoint in US-EAST-1."

@kentik.bsky.social observed a complete drop-off in traffic to this particular AWS service beginning at 06:55 UTC. Thousands of services around the world were impacted.
October 20, 2025 at 2:51 PM
Reposted by Jim Cowie
harvardlil.bsky.social
Our Public Data Project Lead @mollyhardy.bsky.social writes about public data preservation and how it’s complicated by the artificial contemporary distinction between science and the humanities @scholarlykitchen.bsky.social
Guest Post - Rethinking Disciplinary Data Regimes - The Scholarly Kitchen
Between a political policy environment focused on defunding and deleting data collections – an environment in which little can be trusted – and an onslaught of new AI tools that feed indiscriminately ...
scholarlykitchen.sspnet.org
October 8, 2025 at 2:57 PM
Reposted by Jim Cowie
stim3on.bsky.social
Well, forget what I said, turns out you can process marsrover images on the train too :)

There is a super faint smudge where comet 3I/Atlas should be. The new orbit parameters in Stellarium are a pretty good match. 🔭 #3iatlas
October 3, 2025 at 2:53 PM
Reposted by Jim Cowie
jbakcoleman.bsky.social
Open science folks would get a *lot farther* if they spent did a yea long internship in science far outside of their own discipline.

Scoop some poop in the pine barrens and count prions for a bit.
October 3, 2025 at 12:53 AM
Reposted by Jim Cowie
eldomador.bsky.social
Internet down for a second day in Afghanistan as country's first-ever internet blackout continues.

Directive from Taliban govt is to revert to 2G-only service as a way to "prevent immorality."
September 30, 2025 at 11:32 AM
Reposted by Jim Cowie
eldomador.bsky.social
The ongoing internet shutdown in Afghanistan is the outcome of what many feared when the Taliban took over in August 2021.

As of August 2024, we hadn't seen much change with respect to internet service in the country, as I detailed in this blog post:
www.kentik.com/blog/afghani...
September 29, 2025 at 2:58 PM
jimcowie.bsky.social
At long last, AFRINIC can convene a board, hire a CEO, and continue its critically important work. 🙏

www.theregister.com/2025/09/15/a...
AFRINIC has elected a board. Now the hard work begins
: Regional internet registry faces numerous critics and isn’t out of the legal woods
www.theregister.com
September 18, 2025 at 5:41 PM
Reposted by Jim Cowie
eldomador.bsky.social
Beginning at 04:27 UTC today (Sep-15), Starlink suffered a brief but global service outage. This is most widespread service outage since Jul-24 when Starlink was down for more than two hours.

Nearly all AS14593's BGP routes were briefly withdrawn at the time of the incident.
September 15, 2025 at 3:55 PM
Reposted by Jim Cowie
eldomador.bsky.social
Submarine cable cuts in the Red Sea disrupt connectivity between Europe and Asia.

Latency spikes as traffic is re-routed around the broken cables. @kentik.bsky.social's Cloud Latency Map captures the impacts here:
clm.kentik.com?q=4084ef2b1d...
September 7, 2025 at 8:10 PM
Reposted by Jim Cowie
jbakcoleman.bsky.social
Get fucked, textbooks.

Seriously breaking my PhD in ecology and evolutionary biology brain that this happened.

www.nature.com/articles/s41...
September 6, 2025 at 3:08 AM
Reposted by Jim Cowie
jonathanzong.com
Interested in accessibility, visualization, or data ethics? I'm looking for Ph.D. students to join @data-and-design.org at CU Boulder Info Sci in Fall 2026.

data-and-design.org/recruiting/phd
Recruiting Ph.D. Students for Fall 2026
We are an interdisciplinary research group at the University of Colorado Boulder that uses design to understand and reimagine socio-technical systems.
data-and-design.org
September 4, 2025 at 8:01 PM
Reposted by Jim Cowie
eldomador.bsky.social
On Aug 19, China's Great Firewall began blocking most traffic on TCP port 443. This disruption lasted from 16:35 to 17:48 UTC (00:35 to 01:48 Beijing Time).

According to the GFW Report, the outage was caused by the injection of forged TCP RST+ACK packets.
gfw.report/blog/gfw_unc...
August 22, 2025 at 3:15 PM
Reposted by Jim Cowie
tedunderwood.com
This is btw the trajectory you would expect if you believe LLMs are—not an abstract protean principle called “intelligence”—but a cultural technology that animates the verbal machinery already folded up in libraries, allowing it to run by itself.
tedunderwood.com Ted Underwood @tedunderwood.com · Aug 7
This is the world I hope we're getting: One where 1) progress is loosely sigmoid, not because there's any fixed cap, but because the initial overhang of human training data has been exploited, and 2) open models can come close to catching up, because of #1.
A line chart titled “Quality of solutions by some hypothetical metric where human experts are 1.0.” The vertical axis ranges from 0.00 to slightly above 1.00, and the horizontal axis runs from 2015 to 2035. Two sigmoid-shaped curves are shown: a blue “Closed models” line that rises steeply from near zero around 2021 to level off around 2028, and a red “Open models” line that starts similarly low but rises about two years later, leveling off around 2030. Both lines continue to rise rather than leveling entirely flat. The legend is in the lower right corner inside the plot area.
August 20, 2025 at 12:15 PM
jimcowie.bsky.social
"On July 4, 1903, the Manila cable was spliced into the Pacific cable in Hawai‘i, and President Theodore Roosevelt, who was at his home in Oyster Bay, New York, sent the first message to William H. Taft in the Philippines."

doi.org/10.1016/j.ma...
Redirecting
doi.org
August 16, 2025 at 8:53 PM
Reposted by Jim Cowie
tedunderwood.com
I love this opposition, because it really clarifies the difference between assuming language models are abstract intelligences, and observing the roles they actually play in a network of social, language-using animals
timkellogg.me Tim Kellogg @timkellogg.me · Aug 13
instead of AGI we got.. gpt-4o withdrawal

unexpectedly (or maybe expectedly), users formed a psychological bond with 4o and ripping it away seems to have cut them deep

where to go from here?
Sam Altman @sama If you have been following the GPT-5 rollout, one thing you might be noticing is how much of an attachment some people have to specific Al models. It feels different and stronger than the kinds of attachment people have had to previous kinds of technology (and so suddenly deprecating old models that users depended on in their workflows was a mistake). This is something we've been closely tracking for the past year or so but still hasn't gotten much mainstream attention (other than when we released an update to GPT-4o that was too sycophantic). (This is just my current thinking, and not yet an official OpenAl position.) People have used technology including Al in self-destructive ways; if a user is in a mentally fragile state and prone to delusion, we do not want the Al to reinforce that. Most users can keep a clear line between reality and fiction or role-play, but a small percentage cannot. We value user freedom as a core principle, but we also feel responsible as a core principle, but we also feel responsible in how we introduce new technology with new risks. Encouraging delusion in a user that is having trouble telling the difference between reality and fiction is an extreme case and it's pretty clear what to do, but the concerns that worry me most are more subtle. There are going to be a lot of edge cases, and generally we plan to follow the principle of "treat adult users like adults" , which in some cases will include pushing back on users to ensure they are getting what they really want. A lot of people effectively use ChatGPT as a sort of therapist or life coach, even if they wouldn't describe it that way. This can be really good! A lot of people are getting value from it already today. If people are getting good advice, leveling up toward their own goals, and their life satisfaction is increasing over years, we will be proud of making something genuinely helpful, even if they use and rely on ChatGPT a lot. If, on the other hand, users have a relationship with ChatGPT where they think they feel better after talking but they're unknowingly nudged away from their making something genuinely helpful, even it they use and rely on ChatGPT a lot. If, on the other hand, users have a relationship with ChatGPT where they think they feel better after talking but they're unknowingly nudged away from their longer term well-being (however they define it), that's bad. It's also bad, for example, if a user wants to use ChatGPT less and feels like they cannot. I can imagine a future where a lot of people really trust ChatGPT's advice for their most important decisions. Although that could be great, it makes me uneasy. But I expect that it is coming to some degree, and soon billions of people may be talking to an Al in this way. So we (we as in society, but also we as in OpenAl) have to figure out how to make it a big net positive. There are several reasons I think we have a good shot at getting this right. We have much better tech to help us measure how we are doing than previous generations of technology had. For example, our product can talk to users to get a sense for how they are doing with their short-and long-term goals, we can explain sophisticated and nuanced issues to our models, and much more.
August 13, 2025 at 12:33 AM
Reposted by Jim Cowie
tedunderwood.com
The reverse of the bitter lesson is when you spend a month trying to solve a problem with deep learning and then solve it in a day with feature engineering and XGBoost. Let's call it the nauseatingly sweet lesson.
July 19, 2025 at 11:58 AM
Reposted by Jim Cowie
eldomador.bsky.social
New entries in @Starlink's GeoIP file for #India appear the day after the country's space regulator granted a license to launch operations in India.

This was the final regulatory hurdle for the satellite provider to enter the massive South Asian market.

www.reuters.com/world/india/...
July 11, 2025 at 3:24 PM
jimcowie.bsky.social
It took more than a century of suffering on the American frontier before they figured out the "milk sickness" that carried off untold numbers of settlers, including Abraham Lincoln's mother. What a story.

hikersnotebook.blog/flora/other-...
White Snakeroot
Common Name: White Snakeroot, Indian or White Sanicle, Richweed – Snakeroot is used as a common name for several plants without clear attribution. It is generally asserted that this would mean that…
hikersnotebook.blog
July 8, 2025 at 12:58 PM
Reposted by Jim Cowie
eldomador.bsky.social
Tuvalu has became the latest site of a Starlink Community Gateway.

At 20:40 UTC on July 2, 202.2.120.0/21 appeared in the global routing table — the first route originated by Tuvalu Telecom (AS23917) and transited by Starlink (AS14593).

More here:
www.kentik.com/blog/starlin...
July 7, 2025 at 9:06 PM