iiinova — where your digital self keeps evolving

Michael Herf @mherf.bsky.social · 05/10/2026

Why in the world does every journal site now have some kind of AI summarizer in it? If you're skeptical/empirical about it, we don't have proof that AI summarizers work well at this level - they are way too trusting, for instance.

000

Michael Herf @mherf.bsky.social · 24/09/2026

My iPhone can do several teraflops and I can do maybe one. But without a neural interface to it, I am computing tips and negotiating car prices, and thinking about grocery inflation all in my head. You don’t get the benefit of compute for free, it’s only there if the interface to it actually works.

010

Michael Herf @mherf.bsky.social · 08/09/2026

I wonder if the relentless training of LLM frontier models to improve at code/security/math will make them worse at other things? It could be an interesting but constrained rabbit hole.

030

Michael Herf @mherf.bsky.social · 29/08/2026

Two hours per app?! Instagram+Tiktok+Youtube+Netflix and you still spend 8hrs online, til midnight? - a default daily time limit of two hours for minors, with only parents having the ability to opt kids out. - a default block on nighttime use from midnight to 6 a.m. text.npr.org/nx-s1-5947960

text.npr.org

Lights out, Instagram off? The changes to Meta for teens could be a big deal

000

Michael Herf @mherf.bsky.social · 17/06/2026

I hear algae like warmer temperatures and CO2

031

Michael Herf @mherf.bsky.social · 01/06/2026

A major cultural problem with AI coding harnesses is around change control. Just before release, human engineers make the smallest changes possible to fix a bug. They go slowly on purpose. AI harnesses don't have a built-in sense of this - a 2000-line refactor is fine at any time!

020

Michael Herf @mherf.bsky.social · 26/05/2026

Also why do mammals have so many opsins expressed in peripheral tissues? Maybe some is vestigial

110

Michael Herf @mherf.bsky.social · 26/05/2026

Trouble is probably 80% of new content online is AI generated. We will need AIs just to filter out the AI spam.

120

Michael Herf @mherf.bsky.social · 25/05/2026

Wonder if the “obviously AI” text is deliberately written that way precisely so future AIs can exclude it from training

020

Michael Herf @mherf.bsky.social · 24/05/2026

It's a little weird commenting on Reddit now, because in the back of your mind, you're thinking, I'm kind of pre-training the next LLM.

010

Michael Herf @mherf.bsky.social · 20/05/2026

This eats a lot of potential "freemium" OpenAI revenue, because there will be a free Google AI basically forever.

000

Michael Herf @mherf.bsky.social · 19/05/2026

There must be 500 cars per traffic light. Imagine if we put FSD hardware in smart traffic lights, it would save more time and have more safety impact than the same hardware in a small number of cars.

010

Michael Herf @mherf.bsky.social · 18/05/2026

You have to imagine that each conversation has gigabytes of user context, and "utilization" of a big GPU means generating lots of tokens. It does not mean loading all this state just to ask a three word question. But that might be what we (humans) need an AI to do, sometimes.

010

Michael Herf @mherf.bsky.social · 18/05/2026

I think there is an appetite for using AI to make you smarter, but that the cloud models are structurally unable to do it - for them to do a Socratic Dialogue would be a punishingly bad use of a cloud GPU, endless swapping of user state for only a tiny number of billable tokens.

110

Michael Herf @mherf.bsky.social · 17/05/2026

AI through a social lens: 10 years ago we got algorithms that prevented us from hanging out with real people. Now we have new AIs that prevent us from working with real people. I wonder if a society that places less value on individualism would have better defenses against these things.

050

Michael Herf @mherf.bsky.social · 16/05/2026

We even use Naka-Rushton to do tone mapping! It’s rather similar to film and only digital is the linear outlier. I wonder if the new aesthetic that says everything should be as bright as possible pushes us higher in the characteristic curve and loses shadow detail?

111

Michael Herf @mherf.bsky.social · 06/05/2026

www.biorxiv.org/content/bior... Interesting preprint [in mice] says that cone bipolar cell "OFF" (the one shared by rods and cones) is responsible for retinal entrainment. The authors suggest that the fade of sunset could help fine-tune retinal clock timing.

biorxiv.org

022

Michael Herf @mherf.bsky.social · 24/04/2026

AI can score 100% in CMU's 15-112, but... "Last January, many students started to use AI on every assignment, only to drop the class once they began receiving low marks on their quizzes and exams. That semester had the highest number of students who dropped the class since teaching the course."

the-tartan.org

AI’s impact on education at CMU and beyond

With the rise of models such as ChatGPT, Microsoft Copilot, Claude, and Google Gemini, more college students than ever are using generative AI tools to help with their classes.

110

Michael Herf @mherf.bsky.social · 13/04/2026

If there was an important guy who posted on social media a lot but "never sleeps" you might be able to figure out that he has started napping a lot recently.

020

Michael Herf @mherf.bsky.social · 05/03/2026

If you're interested in that one, Tom Wehr's work is also good, though it's mainly in bipolar individuals: www.nature.com/articles/mp2...

nature.com

Bipolar mood cycles and lunar tidal cycles - Molecular Psychiatry

Molecular Psychiatry - Bipolar mood cycles and lunar tidal cycles

010

Michael Herf @mherf.bsky.social · 05/03/2026

"Lunar gravity predicts sleep timing" - bookmarking this lunar rhythms & sleep paper as interesting. It uses a 30-day rhythm but not the 28-day declination rhythm - a claim of "gravity" should probably use the 28-day one, but I think "phase" should follow 30? europepmc.org/article/med/...

europepmc.org

Europe PMCEurope PMC

Europe PMC is an archive of life sciences journal literature.

111

Michael Herf @mherf.bsky.social · 27/02/2026

I think yes, 8AM could be <CBTmin if you've been up in moderately bright light until 4AM. Teens are starting from a late phase already - the amazing thing with this graph is that it is *all of them*. Wonder if teens have a larger CBTmin-DLMO phase angle compared to adults?

000

Michael Herf @mherf.bsky.social · 26/02/2026

Amazing data here - just 100lux of room light while staying up late seems to delay teens enough that 8000lx of morning bright light delays them further:

382

Michael Herf @mherf.bsky.social · 25/01/2026

Here is some code, including a Sprague resampler for upsampling spectra (the CIE standard way), if you're into that! github.com/herf/specwork/

github.com

GitHub - herf/specwork: workspace for spectral calcs (f.luxometer + ecological etc.)

workspace for spectral calcs (f.luxometer + ecological etc.) - herf/specwork

000

Michael Herf @mherf.bsky.social · 25/01/2026

I helped Travis with some moonlight calcs for this paper - our code figures out "how bright compared to moonlight is this light for each species". There is a ton of photometrics and spatial modeling here too: www.sciencedirect.com/science/arti...

sciencedirect.com

Assessing and mapping taxon-specific effects of ecological light pollution for environmental impact analysis

Predicting the adverse effects of light at night is necessary to ensure compliance with laws and regulations that protect species. Using two case stud…

151

Michael Herf @mherf.bsky.social · 03/12/2025

They try to help teens by pulling out the 10% worst of social media… while leaving the five hours a day of not sleeping or socializing, as if that stuff doesn’t matter.

010

Michael Herf @mherf.bsky.social · 03/12/2025

As well, papers that need blue light to penetrate deep into tissues for the mechanism to work... well blue light doesn't do that.

010

Michael Herf @mherf.bsky.social · 03/12/2025

Lots of higher-quality older work that says the closer to green you get, the more you induce sleep/torpor. And also that mice experience "aversion" to blue light and reduce activity. So shining bright blue on mice is not a good model of people or mitochondria. journals.plos.org/plosbiology/...

journals.plos.org

110

Michael Herf @mherf.bsky.social · 04/09/2025

Facial age estimation requires a 5-7 year margin of safety in validation studies, so when the software says you're 23-25 you can be considered 18. Errors are even larger around age 13, so basically useless for distinguishing minors from adults.

000

Michael Herf @mherf.bsky.social · 02/08/2025

👏 Ceiling lamps light up paper but windows light up retinas

011

Michael Herf @mherf.bsky.social · 30/07/2025

illustrating how mDFD can apply directly to a dose-response for a "brighter" light

010

Michael Herf @mherf.bsky.social · 29/07/2025

We proposed a new metric (mDFD) for blue-blocking glasses. It's based on the sensitivity of melanopsin, log-transformed to resemble optical density: tvst.arvojournals.org/article.aspx...

tvst.arvojournals.org

Optimizing the Potential Utility of Blue-Blocking Glasses for Sleep and Circadian Health | TVST | ARVO Journals

250

Reposted by Michael Herf

Helen Branswell 🇨🇦 @helenbranswell.bsky.social · 09/06/2025

This is the big one: Kennedy is firing the entire #ACIP in a move he says is needed to restore confidence in vaccines. "It scares me to think of what’s ahead,” Mike Osterholm of CIDRAP told me. www.statnews.com/2025/06/09/r...

statnews.com

Health secretary RFK Jr. abruptly fires CDC vaccine advisory panel

Health secretary Robert F. Kennedy Jr. has taken the extraordinary step of firing the expert panel that advises the CDC.

22263134

Michael Herf @mherf.bsky.social · 01/06/2025

It's troubling that LLMs do not represent probability distributions *at all* - if you ask a question that should have a normal distribution, it will pick almost the same number every time. Maybe it's the mean, who knows - but it isn't equipped to reason about human populations this way.

000

Reposted by Michael Herf

Peter Gleick @petergleick.bsky.social · 31/05/2025

There are 2 previous historical cases of countries destroying their science and universities, crippling them for decades: Lysenkoism in the USSR and Nazi Germany. The Trump administration will be the 3rd. It's not just budgets but research, institutions, expertise, and training the next generation.

Graphs showing 25 years of budgets for the National Institute of Health, NASA, and the NSF. In all cases, the proposed budget for next year is far, far below any year of the previous quarter century.

449151517827

Michael Herf @mherf.bsky.social · 09/05/2025

Agreeing that we need better names, I would add that it bugs me when people say the human L-cone peak is "red". It looks to me much more like "yellow", and then it's only via L-M opponency that we get the "red" color response?

010

Michael Herf @mherf.bsky.social · 08/05/2025

It would be cool if AIs only ran on solar because then they’d have circadian rhythm.

021

Michael Herf @mherf.bsky.social · 23/04/2025

Astonishing technical feat here, laser projector that tracks eye movements and stimulates cone cells by kind! In just a few milliseconds of latency.

031

Michael Herf @mherf.bsky.social · 13/03/2025

This is a business model problem. Supporting a niche device for a long time is very expensive.

010

Michael Herf @mherf.bsky.social · 12/02/2025

Yet to come, I will post a followup when I have it.

020

Michael Herf @mherf.bsky.social · 12/02/2025

Just received word that ANSI has approved our "flicker" TM, (ANSI/IES TM-39-25), which addresses in delightful detail how flickering lights of all kinds affect people. Congrats to everyone involved with the IES Vision Science Committee.

130

Michael Herf @mherf.bsky.social · 20/01/2025

Same concern as this post: PurpleAir (highly rated by AQMD) and IQAir are reporting 500% the AQI of the single reference sensor. This is the "reference" sensor used by everyone near the Palisades fire, and surely it should not be.

010

Michael Herf @mherf.bsky.social · 12/01/2025

If you’re in an area affected by fires, turning up your car’s fan helps with air quality - cabin air filters are pretty decent.

051