Профиль

Geographically and temporally extensive datasets—informed by theory—are worth their weight in gold. They’re like planting a tree: you’re doing it as much or more for the next generations. LTERs, NEON, and distributed experiments like NutNet, those folks are our tree planters.

Josh Turek has plans to make life better for every Iowan. Now, Iowa, the ball is in your court. Let’s work to get @joshturek4iowa.bsky.social to the U.S. Senate. Turek4Iowa.com

bsky.app/profile/mike...

Hey Lars, I was just at a conference where contributors repeatedly called for ‘starting more long-term datasets’. Two questions in that vein to an expert: How long has this (fantastic) project been going? What has been the biggest challenge in keeping it going? Anything unexpected?

Prediction: The orange guy will turn the newly dead Lindsey Graham into a scapegoat, say he and Bibi misled him into the war (no lie), and use that as his excuse to abandon the conflict.

Sleepy Don dozing off again in public in the middle of the afternoon at Lindsey’s service.

Yes! A writer is a control freak, manipulating every phrase, comma, em-dash, to create an impression in the mind of the consumer. Go on camera? Hoo boy. Without long training and talent, your face betrays you, your brilliant sentences fall muddy and flat, and the editing room can’t save you.

Would Jesus be as famous if he survived?

{begin modest rant} The single most manipulative three word phrase in social media is “Wait for it”. Just for a moment try to imagine the sum total of human existence that has been vampired away by this directive. {end modest rant}

Closer now to 2 out of 3 "key references" are hallucinations. The more specific the question, the more it makes things up, starting with the link, then the title. I wonder it it's default mode is to just rewrite stuff. Or that it's hard to report "that info doesn't exist".

Maybe there is an LLM doppleganger to Reviewer 2. Gemini seems to hallucinate a lot of Key Reference Number 2's.

One of the best tricks I've found for evaluating output from LLM's like Gemini is "include links to key references". The more challenging the question—that is, the reason you're using the LLM for in the first place—the more likely the links don't work, and the whole citation is hallucinated.

A fertilization experiment in a Panama forest shows that invertebrate and fungal species richness increased *most* when fed a *diverse* cocktail of #micronutrients. In contrast, the effects of the "big three" (N, P, and K) were more varied. michaelkaspari.org/wp-content/u...

This makes me intensely happy. Art, Art, and more Art! www.nytimes.com/2026/07/28/n...

Oh fuhhhh.

Wow. How one video can make you happy, then very, very sad, is kind of astounding.

Wiley aside*, publishing with an ESA journal supports a society that recognizes achievement from grad students to old farts, and that actively promotes education and conservation. Publishing with Springer/Nature? You will make some stockholders happy. *I know, it’s a complicated world.

Just a funny funny human. May his memory be a blessing.

ICE claimed Lorenzo Salgado Araujo had drugs in his van. The lab results are back: That was false. This was a man who spent decades breaking his back on construction sites to provide for his family. ICE fatally shot him, then tried to smear his name.

civileats.com/2026/07/27/w... “What aggravated the legislature was not so much the data collection itself, but that the data was so easily accessible by the public,” Jones said. “That discredits what they’re hearing from politicians about the nutrient strategy working.”