cleandreams

My son has ADHD and I never knew what it meant but this year I made him get coaching for it and I made him get meds. He was living at home for free rent and the condition of staying was to address this. My experience is that getting treatment for this condition helped him a lot. The coach really helped, the meds really helped. He is full of zest for the work that lies ahead. He's off on his own now and while I think he'll have stumbles it's basically very positive.

One thing about ADHD coaching. The coach is a consistent person who helps frame a positive attitude with agency and skills and this takes you away from negative self talk. It's good. I think ADHD folks need this positive support.

There is more to this story. I have heard Harvard (and its ilk) described as a smallish university attached to a massive investment fund. One of the criteria for admission is how much you and your family will be "good for" donations to the university. They much prefer students from rich families, even if they are the children of disagreeable dictators.

No. Briahna Joy Gray and a guest whose name I don't remember.

We'll be cleaning up tech debt from over-reliance on AI.

I was listening to a podcast where the host and the guest were both Black and both had been to Harvard. This was before the Supreme Court ruling. They joked that it was well known that the Black students at Harvard who were descendants of American slaves were about 10% of the Black student body at Harvard. There were a lot of children of Nigerians and 3rd world tycoons and such.

I think there is a good argument for a help up for people whose communities are still impacted by the history of American slavery (and all its ills) but giving these slots to children of the wealthy and often immigrants does not feel right.

what is missing from this is the issue of power. tech companies interpret the rise of AI as empowering management and disempowering workers.

It's the American way to disempower workers. Tech people used to have a moat called expertise and management believes AI has made that obsolete.

Eventually workers in America will need to fight for power, probably through unions. That is the time honored way. There will be a lot of degradation and impoverishment on the way there I think. Fasten your seatbelts.

I was walking through a startup neighborhood in San Francisco the other day and I encountered several telephone poles which were posted with advertisements for software engineering jobs. These were not generic or scam advertisements. This was a particular startup, looking for software engineers. What is old is new again.

Shout out to American Giant. I have like 15 t-shirts from them, some are ten years old and they look new. Buying good stuff saves money in the long run!

Not just many foreign militaries. Our military. General Dan Caine by report advised Trump about the negative consequences of this action. MAGA elected a fantasist and narcissist and we will all bear the consequences. I was no fan of Kamala but the appalling limitations of Trump are of a different order and it is a disgrace that our voters by majority elected him.

Wonderful to hear. Her long Covid was heartbreaking (saw the videos). I hope she gets stronger and stronger! Welcome back!

Private car ownership is better everyday for suburbs and rural areas but in cities that is not true. Public transit can improve downtown access and reduce congestion. You need some density for transit.

San Francisco put in some bus only lanes and those routes have greatly improved bus speed and ontime performance.

This is not Facebook but I have a young friend who gets her news on Epstein from tiktok. She is convinced they were eating babies. I do worry (a little) that she will go over the conspiracy theories deep end. I told her tiktok news was bad "mental hygiene" but she didn't get it.

My son -- early 30's -- is impacted by how everyone's life seems much better than his. I think it's a real issue for young people.

To be fair SF has had incomprehensible (to normies) billboards since at least the early 90's.

I have a friend who was put on a GLP-1 and had an astonishing reduction in arthritis pain.

The prospect of going the last mile to fix the remaining problems reminds me of the old joke:

"The first 90 percent of the code accounts for the first 90 percent of the development time. The remaining 10 percent of the code accounts for the other 90 percent of the development time."

I'm anxious about code quality in critical infrastructure in 5 years or so.

Also my mastery of code starts with design and implementation that results in deep, intuitive understanding. Then I can do good code reviews and fix bugs fast fast fast.

Now engineers leap from AI assisted or even dominated implementation to code reviews. Lots of reading code without that deep level of mastery. With this approach I have less confidence in the humans who are in the loop.

If you can extract a whole copyrighted book from a large language model, what is the point in copywriting your book?

What I find missing in the vibe coding discussions is precisely what we can't know, which is how this will work out over time in practice. A couple concerning data points I gleaned recently (sorry no references): - the LOC per PR is going up. - the number of bugs per LOC is going up.

I wonder whether 'enshittification' will hit the installed software base hard in the future and the software engineers tasked to fix it will not have been brought through the rigors of designing and implementing complex systems on their own, without LLMs.

This doesn't address the degradation that may happen when LLMs are increasingly trained on LLM output.

BTW I used to work in AI and I don't think scaling is a solution here.

For the first time I have begun to doubt Microsoft's chosen course. (I am a retired MS principal engineer.) Their integration of copilot shows all the taste and good tradeoff choices of Teams but to far greater consequence. Copilot is irritating. MS dependence on OpenAI may well become dicey because that company is going to be more impacted by the popping of the AI bubble than any other large player. I've read that MS can "simply" replace ChatGPT by rolling their own -- maybe they can. I wouldn't bet the company on it. Is google going to be eager to license Gemini? Why would they?

This is great. Thanks for sharing. Should be a book someday.

I'm amazed no one brings up the obvious: the need for reactions that are reliable at high speed. There is no way I will trust my tuckus to a freeway driving Waymo for a couple of years.

The judge IIRC found that training models using copyrighted materials was fair use. I disagree. Furthermore this will be a problem for anyone who generates text for a living. Eventually LLMs will undercut web, news, and book publishing because LLMs capture the value and don't pay for it. The ecosystem will be harmed.

The only problem the judge found here was training on pirated texts.

I think the universities made a mistake in becoming too culturally left wing. The faculty (and students) in the humanities in particular are far to the left of the political mainstream. (I am on the left though more focused on labor rights and good jobs than identity politics.)

The attack on the universities is fueled by this divergence, now that the right is firmly in power. This will just hurt the country in the long run. There was so much group think and silencing happening on the left over the last decade. It seems now to have been self-destructive.

Less than 10 years ago but not recent.

My base salary was fine but the magic was in the stock.

I got a payout on acquisition by a FAANG+ (as first employee). It was only 300K but I put 50K of that into Nvidia. Actually I invested all my payout from my startup stock into tech stocks. And I got a terrific golden handcuffs deal.

After that I could afford to retire and I did.

At first it seemed LLM's were perfect assists for coding because they are trained on text and generate text. But code isn't typical text. It's basically a machine that requires a very high degree of precision and accuracy. Seen this way, LLM's are suited for coding only at specific stages -- to generate something like boiler plate, to brainstorm, to evaluate diverse approaches, identify missing tests. Anything that ties LLMs to actual code implementation is asking for trouble in my view.

The weird thing about AI is that it doesn't learn over time but just in context. It doesn't get better the way a 12 year old learning to play the saxophone gets better.

But using it heavily has a corollary effect: engineers learn less as a result of their dependence on it.

Less learning all around equals enshittification. Really not looking forward to this.

A paper to make the teachers I know weep.