HN user

ckemere

1,388 karma

Professor of Electrical and Computer Engineering {Machine Learning, Neuroengineering Technology, Learning and Memory}

Posts19
Comments315
View on HN
arstechnica.com 3mo ago

To teach in the era of ChatGPT is to know pain

ckemere
57pts25
www.washingtonpost.com 10mo ago

Hegseth orders all generals to a mysterious, in-person meeting

ckemere
19pts6
molochinations.substack.com 1y ago

Switching from Manager to IC

ckemere
3pts0
news.ycombinator.com 1y ago

Has anyone else found Google's AI overview to be oddly error prone?

ckemere
43pts39
www.energy.gov 1y ago

Dept of Energy cuts indirect rates to universities

ckemere
3pts2
www.theguardian.com 1y ago

Defying court order, Venezuelans deported to El Salvador under Alien Enemies Act

ckemere
4pts1
www.thetransmitter.org 1y ago

NIH grant review (secretly) paused

ckemere
9pts4
mathstodon.xyz 1y ago

Complex dynamics require complex solutions

ckemere
214pts144
www.wsj.com 1y ago

Happiest Baby Angers Customers

ckemere
7pts2
news.ycombinator.com 1y ago

Ask HN: Marketplaces for Embedded/FPGA Developers

ckemere
43pts33
onlinelibrary.wiley.com 3y ago

ML-Derived Facial Characteristics and Financial Analyst Performance

ckemere
1pts0
news.ycombinator.com 3y ago

Ask HN: Has anyone else had trouble with Google Takeout

ckemere
3pts0
www.washingtonpost.com 3y ago

Small business prospers removing confederate statues

ckemere
1pts0
peaksalvation.com 3y ago

Former tech executive's experience in Amazon warehouse

ckemere
2pts0
www.washingtonpost.com 4y ago

Imagine I-95 catastrophe with EVs

ckemere
9pts6
news.ycombinator.com 5y ago

Ask HN: Why can't the NBA have promotion and relegation?

ckemere
4pts7
newrepublic.com 5y ago

Against the Consensus Approach to History

ckemere
2pts0
coronavirus.rice.edu 6y ago

Rice University Reopening Plan

ckemere
2pts0
www.theatlantic.com 6y ago

My Family's Slave [2017]

ckemere
2pts0

Perhaps a huge tell about national strategy is the fact that the owner has $10s of millions to loan to the company? US economic structure in post WWII era has increasingly focused on return on capital (and value extraction). How can that compete in long term with an economy which prioritizes reinvestment *in industry*?

I found this article frustrating. In my experience, the students who participate in the “co-creation of a contract” would likely also have done their best to comply with Professor induced rules. The challenge is the 10-20% who, stressed and overworked, will be just as likely to ignore this contract as any other. I would have liked to understand how the situation is changed compared to no rules at all…

Comparing with China, I’d be curious about whether their companies face the same return-on-capital forces that US (and to a lesser extent western) companies do. So when we outsource to them, it’s to a vendor who is not needing to promised 10%+ returns. As I understand it, this is because the central government exerts pressure on banks to make loans at low rates, so that capital can be raised more easily for targeted industries.

I agree that onshoring is likely impossible without some sort of contextual reimagining of US industry.

Training data of task completion. See, e.g., robots doing backflips. Presumably there’s an optimal robot for gymnastics but if you start with humanoid form you can train based on many videos of human movement. The alternative - world model sim with physics and loss functions- perhaps ends up being too unconstrained when you add in optimization of the robot form…

Lots of complaints about corruption. It’s worth imagining how you would run a very low income developing country. Remittances and tariffs are the easiest items to tax for revenue. I’d love suggestions for better alternatives…

Curious about the design choice. Why not use the TI parts with integrated microcontrollers rather than two separate chips? Or even a FPGA with integrated ARM9 like the Zynq family?

The obvious question I would have asked: given the concern that this may not be ethical if the brains are still “alive” AND the concern that a brain separated from the body probably doesn’t function these same, why wouldn’t we test things in living monkeys (instead of mice)???

It seems that the likelihood is high that the right animal model would yield superior data???

To the people who are thinking about other languages, "Lexical decoding" - recognition of a word during reading was the strongest predictor of reading compression (as opposed to phonological).

Restating the highlighted result: Gc ("Comprehension-Knowledge") had the strongest effect on both lexical and phonological decoding. Knowing a word makes it easiest to comprehend when reading. This is probably completely obvious, but the broader point is that rich conversations with students that involve teaching them lots of words will improve their reading.

Only partially supported interpretation/application - All this business about phonics will only take you so far if the adults in a kids life (including their teachers) are not talking to them richly about a lot of stuff. Asking teachers to do a lot of rote repetition risks cutting out the really important part of school where students are actually building vocabulary. Teachers that use/teach large vocabularies may be unexpectedly more effective at teaching reading.

I’d love to take that bet. My deed (in Texas) states that my lot is subject to the rules of the subdivision which include a number of zoning style restrictions. (They’re called “deed restrictions” and are very common AFAIK.)

The subdivision rules are changeable only with a supermajority vote. I believe the city (Houston in my case) is prohibited by the state from unilaterally changing them.

(I wouldn’t mind more free property rights!!! I find TX “liberty” is often biased towards $$$)

The title really should specify “to teach remotely”. And I think more broadly the context is the dream that widespread internet would make it easier to educate people “at scale” (meaning for less money per student).

So maybe the real question is why we ever expected “teaching at scale” to be effective.

I think that it’s quite clear that for an individual, curious student, the ability to use modern LLMs probably makes the ability to be 1-1 tutored (by a human!) cheaper/better. But I don’t think anyone claims that watching random videos on the internet will be as effective for LeBron James as having a personal trainer focused on him.

It seems like the overriding issue is to understand whether students need to take courses they’re not interested in. If the answer is yes, perhaps we need find ways of having these topics be taught by tutorial…

I think that there are stories but unreported for good reason. I heard from a pastor friend that he showed up to a house where one of his congregants was inside (with ICE outside), and basically asked them for mercy, and the supervisor was like “Actually, there is lightning in the area so we’ll need to go away but we will come back later.”

So not complete rebellion, but a little bit of humanity. Viktor Frankl talks about how there were some “good” guards at the concentration camp.

I’m quite curious about recruiting. I have only a N=1 observation, the kid who takes the orders at my favorite burrito place. He had been hyped about the Marines for two years, pre-enlisted at 16, just waiting to graduate from HS this spring. I didn’t see him for a few months, but in November it had all changed. “They are hostile to people like me.” (He’s of Mexican descent.)

A lot has happened since June of last year. Let’s not forget National Guard deployments to cities and threats of active duty / insurrection act. Threats of sending the army to fight cartels. I think the current situation is just an extension of craziness that would give anyone except the hardest of core supporters pause???

Hot take - the private doctors tell you they are great but the public doctors can often be spectacular because their motives are not primarily economic.

For example, the absolute best diagnosticians in Houston are at the public hospital primarily serving Medicaid and Harris Health patients. Super evidence based, order tests for differential diagnosis not to make $$. Passionate about what they do. In a unexplained emergency my doctor friends would go there to be diagnosed and then the fancy privates to be treated.

Potential counterpoint. Is it possible that one challenge is the lack of expertise in government? I think it’s clear that most novel permitting situations involve one expert party (who want the permit but are potentially motivated to not report downsides) but the other party (the regulator) has to either develop their own expertise or say “no”/“wait”.

I was unimpressed by the situation described. It seems that existing injection wells often have all sorts of negative consequences that are avoided by bankruptcy. I suspect more “no”/“waits” in the past might have been reasonable