Though a local model I'm running (gemma-3-27b-it; https://huggingface.co/lmstudio-community/gemma-3-27b-it-GGU...) just told me various correct sounding bits about his history with alcohol (correctly citing his alma mater and first wife), but threw in:
"Sobriety & AA: Newman got sober in 1964 and remained so for the rest of his life."
Which doesn't check out. And it includes plausible but completely hallucinated URLs (as well as a valid biography.com URL that completely omits information about alcohol.)
I will say this: if you're not happy with your current prescription, there are ways to get a more intense workup and better outcomes by going to an academic optometry center. In my case, I went to the New England College of Optometry and got prescribed a special type of contact lenses ("scleral" lenses) which have been a major quality of life enhancement.
They're expensive, there was a learning curve for getting them on correctly, and it took several followup appointments to get the correct fit from the manufacturer, but I can wear the lenses almost all day and they give me clear, sharp, 20/20 vision.
Also, when I'm wearing them I need reading glasses to read up close--my uncorrected vision actually compensates for my slight age related nearsightedness. But my vision is so much better I don't mind at all!
The back story is that I had lifelong astigmatism and 2 eyes with different powers (one more farsighted than the other one) which led to some mild amblyopia (lazy eye) that I've had since childhood. My vision wasn't "that bad" so I got by without using my glasses for a long time. But when I tried using my several year old prescription glasses I found that presbyopia (that age related inability to focus on anything up close) made the glasses almost useless for reading.
Even though I'm a dev who looks at screens all day, I didn't think I minded, but I noticed in recent years that my appetite for reading books had disappeared was partly due to noticeable eye strain, but also due to generalized eye fatigue that I wasn't really acknowledging. I also had to sit up front in meeting rooms to follow along with anything projected on the screen, which was annoying.
A colleague mentioned the book Fixing My Gaze (https://www.google.com/books/edition/Fixing_My_Gaze/Ul16tPVk...) and I bought it. It's partly a personal narrative by a neuroscientist who was stereoblind and taught herself to develop stereo vision in middle age (she was profiled by Oliver Sacks at one point). But it's also a history of research optometry, which focuses on refractive vision correction and visual processing (as distinct from eye diseases) and which I barely even knew was a thing. Which led me to NECO and my big quality of life improvement!
Should be fixed soon! We decided at 6pm on Friday to reduce biomedical research centers' budgets by billions of dollars ($4B in Massachusetts alone), effective Monday. So in the future China will have to train its own scientists.
After I installed a high-COP, cold-climate heat pump in my Boston area house in 2020, I figured the wintertime running cost was about half again the cost of gas heat. However, the cost was about the same as that of running the somewhat ancient oil furnace the house came with.
Electricity prices in MA are high and are controlled by the marginal cost of gas to run generators, even though we have a significant fraction of renewable, hydro, and nuclear generation. Gas prices are also high because we're at the far end of the east coast natural gas pipeline network.
It's interesting to think about the whole chain -- burning gas in a high efficiency condensing furnace at home approaches 100% efficiency of conversion from chemical to heat energy.
Whereas utility generators that convert heat to electricity are upper bounded by the second law to ~60% efficiency, and then you have transmission losses on top of that. But you win roughly all that lossage back because your heat pump can pump ~3 units of outdoor heat energy into your house for every ~1 unit of electrical energy it consumes.
Add transmission costs and unfortunately heat pumps are more expensive here for now. But CO2 wise of course it still wins because of that renewal and nuclear share I talked about
More importantly, I can also install solar and start getting some energy "for free" (obviously, much more so in the summer than in the winter). And over time of course our renewable share will go up.
I worked there until 10 years ago and it was quite rewarding. (In fact this interactive uses some of the "lab" framework code I worked on.)
The founder was inspired by Concord (Massachusetts)'s history of progressive educational experimentation, going back to the transcendentalists (Emerson, Thoreau).
They're based in Concord but they also have a Bay Area office, after merging with the nonprofit org that built CODAP (https://codap.concord.org/)
I'd rephrase that as "the author is the author, Stephen Johnson, who is also one of the key people..."
I've read many of his books over the last 20 some years (and even watched a PBS documentary series he hosted). I was aware via his Substack that he was collaborating somehow with the NotebookLM team. But I was rather startled when he demoed NotebookLM at a Google all hands meeting a few weeks ago! Apparently he's a full time product manager now.
Fortunately by last year the this Café (GE) double oven induction range was available here in the US: https://www.cafeappliances.com/appliance/Cafe-30-Smart-Slide... I have a few quibbles (mainly, that only one of the burners is properly sized for a 12" skillet) but overall I like it.
I don't mind the touch buttons for operating the oven and timers--in fact, they're nice and easy to clean (with a handy "lock screen" feature so you can spray and wipe down the front panel without everything going nuts) but I'm pretty sure trying to fine tune the burner settings using a touch slider while keeping an eye on multiple pans would have driven me nuts. I also have haven't had problems with the knobs getting dirty or being hard to wipe down if they do, to address a point raised in another reply.
Price splits the difference between the entry level ranges and the snobby brands (Miele, Thermador, etc).
From a quick skim: what's interesting about this study from an HN perspective is that they used Alphafold (or more specifically Alphafold-Multimer: https://www.biorxiv.org/content/10.1101/2021.10.04.463034v2....) to screen more than 1400 different proteins that were likely to be present on zebrafish sperm to see which ones might bind to other known sperm fertility factors.
Lo, and behold, they found a protein (i.e, the product of a particular zebrafish gene) that Alphafold-Multimer predicted would bind to two of the known sperm factors. And it turned out to be a kind of missing link: the three sperm proteins together were predicted form a stable structure. And, that structure ("complex") sticks to the only egg protein known to be required for fertilization! (Where all of this was first predicted using Alphafold-Multimer, then experimentally confirmed to some degree.)
Not only that, it turns out human versions ("orthologs") of these three sperm proteins exist, and their experimental evidence at least suggests that they stick together, forming a complex as well. Which presumably sticks to some human egg protein. Pretty neat.
Why this matters: Consider. 20 years ago, I briefly worked for a lab that used genetics to study fertilization in C. elegans (fast breeding, millimeter-long worms with a lot of infrastructure in place for scientific study). Sure, we were studying worms, but the PI had a personal interest in (in)fertility, and it was his long bet that fundamental research would help medicine solve infertility.
Now it looks like the bet is showing promise of paying off: back then, there didn't seem to be any vertebrate equivalents of the worm genes we found. Maybe worm fertilization was just too far removed. But the top "related article" is from my old lab (https://www.cell.com/current-biology/fulltext/S0960-9822(23)...) and the abstract points out that several worm genes they and related labs found are, in fact, equivalents of the vertebrate genes discussed in TFA! So progress accelerates.
As an engineer who has worked on systems that handle sensitive data, it seems straightforwardly to me to be a statement about:
1. ACLs
2. The systems that provision those ACLs
3. The policies that determine the rules those systems follow.
In other words, the model training batch job might run as a system user that has access to data annotated as 'interactions' (at timestamp T1 user U1 joined channel C1, at timestamp T2 user U2 ran a query that got 137 results), but no access to data annotated as 'content', like (certainly) message text or (probably) the text of users' queries. An RPC from the training job attempting to retrieve such content would be denied, just the same as if somebody tried to access someone else's DMs without being logged in as them.
As a general rule in a big company, you the engineer or product manager don't get to decide what the ACLs will look like no matter how much you might feel like it. You request access for your batch job from some kind of system that provisions it. In turn the humans who decide how that system work obey the policies set out by the company.
It's not unlike a bank teller who handles your account number. You generally trust them not to transfer your money to their personal account on the sly while they're tapping away at the terminal--not necessarily because they're law abiding citizens who want to keep their job, but because the bank doesn't make it possible and/or would find out. (A mom and pop bank might not be able to make the same guarantee, but Bank of America does.) [*]
In the same vein, this is a statement that their system doesn't make it possible for some Slack PM to jack their team's OKRs by secretly training on customer data that other teams don't use, just because that particular PM felt like ignoring the policy.
[*] Not a perfect analogy, because a bank teller is like a Slack customer service agent who might, presumably after asking for your consent, be able to access messages on your behalf. But in practice I doubt there's a way for an employee to use their personal, probably very time-limited access to funnel that data to a model training job. And at a certain level of maturity a company (hopefully) also no longer makes it possible for a human employee to train a model in a random notebook using whatever personal data access they have been granted and then deploy that same model to prod. Startups might work that way, though.
There was a Soofa sign near my house across the river in Somerville that (now that I think about it) I haven't seen for a while now. The branding wasn't quite as prominent as the Brookline Bank sponsored one in the photo. As I recall it displayed a vaguely useful calendar of local and city events, plus a question you could reply to on Twitter so they could show "engagement".
I'm not surprised nor especially troubled that the sign gathers pseudonymized MAC addresses in hourly buckets. In the last several years there's a mini-trend of startups attempting to provide more or less anonymized smartphone traffic data to cities and towns for urban planning purposes.
In theory this is good! Ideally it helps city hall be more data driven and see things that might not filter up to city hall, in a "pave the cowpaths" way (E.g,. do we need a new circulator shuttle stop? What's happening that one weekend in May that drives so much foot traffic and that we're not aware of at city hall, and should send a police detail to control that intersection? Oh, that brewpub has an annual event we didn't know about that blew up on Instagram.)
In practice I think the problem is the wins tend to be minimal compared to the effort involved.
All that said, I don't love the conspiratorial, low-trust assumptions you're encouraged to make by the bare statement "They’re collecting data from your cell phone."
But without privacy regulations I suppose that's where things will inevitably go -- people will assume a priori that "data collection" is itself threatening. (I certainly foresee a lot of rich retirees agitating to cancel the contract at the next Brookline town meeting.) So I wonder if the main benefit of better privacy regulation in the US would be preventing further deterioration of the basic trust that allows the "collective intelligence" vision of the 00's to come to fruition.
Fun! This is the kind of thing I got back into programming for around 2010. Just to toot my own (and my former employer's) horn, I worked on a bunch of the underlying models behind the Concord Consortium activities here: https://learn.concord.org/
They're aimed more at middle and high school, are "curriculum aligned" and developed into classroom-ready form by professional educators, and then Concord researches how well they work in actual classrooms.
We had been using a Java app (and applet) called Molecular Workbench for some of these, but in 2012 we got a Google.org grant to reimplement the same concept but natively running in browsers.
It traces the story of a Renaissance humanist who tracked down and translated the Epicurean philosopher/poet Lucretius' De Rerem Natura, which Greenblatt describes as portraying a strikingly modern way of seeing the world.
In particular Lucretius and the Epicureans denied the existence of supernatural causes, were opposed to religious fear, and posited the ideas of atomism and biological evolution. Of course they're better known for their approach to living life, which Greenblatt shows is more sophisticated than sometimes caricatured, and which he portrays as a breath of fresh air compared to the oppressive moralism and hypocrisy of the Church at the time. (Jefferson and many of the American Founders described themselves as Epicureans.)
He goes on to imply that Epicureanism was influential and widespread in the ancient world but suppressed by the early Church, so that we now know little of it.
Anyone, one of the tantalizing parts of the book is where he describes the carbonized and unreadable Herculaneum scrolls, since they were the private library of a wealthy patron of the Epicureans. I think he thinks being able to read the scrolls will really change our understanding of the ancient world.
And remember: if they hadn't been carbonized, they would have crumbled to dust. That's why we only have the texts that managed to get copied. (Anthony Doerr's Cloud Cuckoo Land is a novel about the survival and 21st century rediscovery of an imaginary Greek play, and ... I'll let you read it yourself - https://www.anthonydoerr.com/books/cloud-cuckoo-land)
(Apologies for any errors above, as basically all I know about this subject is what I read in the book!)
Ah, I worded that badly. Ultimately the pilot is the decider as you say. I was getting at the idea that (so I am told) the day to day experience of piloting is very much about working within a regimented system.
I can think of many reasons it's difficult, though:
* The fact that accident investigations are designed to figure out which driver was at fault, for insurance purposes
* The fact that road designers are shielded from liability (good, right?), but only if they can show they followed the standard design handbooks, which not only fail to prioritize safety over traffic flow but also don't require or even contemplate performing a thorough case-by-case investigation into how to prevent a certain type of accident from happening again. (Crash Analysis Studio tries to demonstrate how to do this.)
* The sheer scale of the changes to roads that would be needed. In the States, access of hundreds of thousands of businesses and probably millions of homes depend on roads that are supposed to get people from A to B fast but also have lots of access points for businesses on them, which is like just like mixing taxiing and holding with takeoff and landing, but without the air traffic controllers. And don't get me started about unprotected lefts across multiple lanes of traffic.
* The lack of buy-in from the citizenry for enforcing professional-like standards on drivers. Despite the blameless culture that helps identify flaws in the system, air traffic controllers, pilots, maintenance crews, train engineers, etc also know that their job depends on making a sincere effort to follow the rules that are written, and in most cases their professional identity is tied up with following the rules. (I've heard that to be a pilot you really have to be comfortable with doing what you're told, all day long.) There's really no obvious way to get the masses of drivers to think and feel that way about their driving, and of course we can't just sanction our way to compliance because people need their licenses to go about their lives. And of course usually there is no reasonably alternative to driving yourself where you need to go.
You might be interested in author #7. Some guy named Dario something.