Have you considered Yu Ming, the language immersion charter school? You wouldn't need to move, you wouldn't need to pay, and 88℅ of students meet or exceed state standards for math.
(There are folks working at SFUSD for whom Yu Ming was their top choice of school for their kids.)
I grew up in the UK. Electric kettles are more common there than they are in the US because:
- People drink at all times of day, and
- Household power sockets can deliver a lot of power (a typical $30 kettle in the UK is rated at 3kW vs. US where 1.8kW is normal even for more expensive ones)
Anyway, before kettles became cordless[0], they all had detachable leads. And there was a standard. So you could use any 'kettle lead' with any brand or model of kettle.
To this day I still use 'kettle lead' to refer to the type of cable used to power a desktop PC.
[0] the kettle itself has no cord, but the base has a hard-wired cord
One of the best moka pots I owned is the Bialetti Elettrika [1] which comes exactly with a detachable lead like the one you mention. But for me, coming from outside of the UK and never experienced the kettle lead situation, it has always been the opposite, the pot has a ‘computer lead’ :)
Developers can update Claude’s instructions mid-task without breaking the prompt cache or routing the update through a user turn. This can be used in a given harness to update permissions, token budgets, or environment context as an agent runs.
Does this means the instructions are no longer just something in the early part of the conversation? (If they were, changing them would invalidate the KV cache. no?)
Perhaps they trained it with a new special system instruction token that is specifically trained to produce the same result as changing the system prompt, but is inserted into the prompt mid-conversation?
"In 11th grade, the most relevant grade relating to college readiness, 30.5% of students met or exceeded math learning standards. Of these, nearly half exceeded the learning standard — marking them as likely to be the best prepared for a college STEM major."
Politicians in California want the ethnic mix of students at public universities to reflect the ethnic mix of the state population. They cannot achieve this goal if colleges use academic preparedness as the main factor in admissions:
> Politicians in California want the ethnic mix of students at public universities to reflect the ethnic mix of the state population. They cannot achieve this goal if colleges use academic preparedness as the main factor in admissions
That's a satisfiable goal, but it means they have to accept that they need extremely broad and deep remedial courses, and they need to treat admissions more or less like a community college does. Is that what they're looking to do? What are their goals regarding the existing community college system?
The politicians don't accept that students need remedial courses.
They passed AB705 and AB1705 to prevent a community college from putting someone in a remedial classes unless it had very strong evidence they wouldn't be able to pass a regular (transfer-level) class.
So if you go to a community college and intend to study for a STEM degree, you'll be placed in a calculus class.
> They passed AB705 and AB1705 to prevent a community college from putting someone in a remedial classes unless it had very strong evidence they wouldn't be able to pass a regular (transfer-level) class.
Why? Did someone make students graduating from high schools who go on to need remedial courses into some kind of metric for the high schools, and the politicians are trying to "solve" the schools' low scores on this metric by cheating?
I'm sorry, I was not prepared for how insane this is. It's super late, so I'm going to need to do it later, but I guess I should go look up those bills.
While it seems obvious that some students should be redirected to remedial classes, the evidence is that very few students made it past those courses. IOW, the obvious solution wasn't working.
Being an engineer, my instinct would be to fix those classes so that they did work. However, legislators think at a different level and reasoned that the remedial classes constituted a false promise that costs students dearly (time, money, hopes, and dreams).
I have been really puzzled by this situation and haven't been able to stop thinking about it since last night. I spent a long time chatting with an LLM and reading articles published by community college instructors and university faculty in California, and looking for examples in other states, the research (in California and other contexts) that examine the related problems with remedial courses.
I've learned a lot, but the one example I want to raise is one I learned about how some other community colleges have addressed the same problem. In some states that where the solution wasn't mandated or constrained by legislation, schools replaced their conventional placement test and remedial courses track with repeatable, low-stakes testing. When you fail the test, it points out where you were weak and directs you to study material, and then you can study only the parts you struggled with and retake it as soon as you want, as many times as you want, for free. If you fail repeatedly you're offered a kind of integrated online course that is self-paced rather than a fixed semester length and has a really favorable class size (15 students, 2 instructors). It's sold as a service community colleges can buy into, and I really know nothing about it, so I don't want to name the vendor. I don't know if their particular tests are actually good, of if their streamlined course recapitulate any of the failings of conventional remedial courses.
But the general outline seems... pretty good, right? It isn't expensive for the students, it isn't a lengthy detour, and it doesn't work by lowering standards or potentially fraudulently promoting unprepared students (which I imagine adjunct professors at community colleges are systemically pressured to do at institutions where administrators care about their pass rates).
I'm not sure if "Come back as soon as you're ready, here's where you struggled, here's where to get extra support if you need it, all of this is free" should be considered fixing the remedial courses or bypassing them, but it seems doable and like it addresses the time, money, and stigmatization/discouragement problems with old-school remedial tracks.
Anyway I hope California can get more creative here and try to get serious about measuring success (i.e., actually do more testing of learning outcomes when they make changes like this, instead of just looking at course completion rates). It seems like a solvable problem.
Maybe I'm getting old, but I don't see the appeal of reading on an eink device that's smaller than my phone, which I'm always carrying. Maybe if I'm reading outside in sunlight rather than in bed? Or if I'm worried about getting distracted by a FB/X notification?
Different person, but I bought a set for me and my wife on a whim because they’re so cheap, and found I adore the little thing. I have a public transit commute to and from work. Since getting it, I’ve spent my commutes reading books I’ve meant to get around to.
I have a Kobo I keep at home. I love it, but don’t want to risk breaking it while carrying it around in my backpack, and it’s too big to comfortably hold on a crowded BART (let alone to dig around in my bag to get it out and put it away). The X4 is always in my pants pocket during the commute and small enough to break out wherever I am. Also, it’s small enough to not feel fragile, and cheap enough that it wouldn’t be devastating if I broke it anyway.
The appeal is that it is a better reading screen than a phone, while being the same size as a phone. It means that I can take a book anywhere without having to bring a bag
Literally nothing you can lift requires a bag, yet many people choose to carry their possessions in a bag when they are traveling because they find it fairly convenient to use their hands for things.
For anyone reading Dan Meyer's blog for the first time, I'd highly recommend you search the archives for an article about a topic you know well. And use that to help you decide how much to trust Dan's opinion on other stuff, and whether his perspective aligns with yours.
But there's no evidence that he's tested it with a student in the target market!
He does link to article by Michael Pershan, who did kick the tires, but again, he didn't use it as intended over any reasonable period of time.
Based on their credentials, both Dan and Michael are people you'd hope you could trust if you want to learn about how to teach your kid math. Dan did his PhD at Stanford (under Dr Boaler) and Michael has been a math teacher for years, and has written at least one good book on the topic.
But they each seem to have a knee jerk reaction to anything that hints that it might replace human teachers, or anything that's only useful to a subset of students.
reply