yes for sure! And what does it mean to be secure? I think agents complicate even this basic question.
yes for sure! And what does it mean to be secure? I think agents complicate even this basic question.
While we do that, the nature of software engineering will change, most likely merging with the product manager role.
This is my prediction. Software engineers also are subject to Jevons Paradox. As they get "cheaper" per unit work, they will get used more.
There are probably more realizations like this we're going to have: vast swaths of changes that need to be happen, more than we can conceptualize right now. We'll do them with ever greater efficiency, but we will do them.
Software engineering isn't dead, at least not yet. We have to do everything we did before, plus make everything friendly to agents.
There's a ton of work to do to get agents to be useful, and this is just one of the things we're realizing in the age of AI.
Some things I see that need to be done: universal APIs, agent auth, payment infrastructure, and much more.
OpenClaw is fighting a ridiculous battle to be useful against our current software, none of which was designed for agentic use. If you've used it, you know just how unpleasant, confusing, and dangerous the setup is.
I'm not so sure software engineering is a doomed profession.
Yes, companies can do work with far fewer software engineers these days. But using OpenClaw makes me realize just how much work there is to do.
#emacs org-mode is, IMHO, the best system for not only tracking your work, but also tracking and assigning your agent's tasks. But it has a significant flaw that I think my ekg package can address. My video on this:
www.youtube.com/watch?v=CleN...
Yeah this is true but it's also why experienced devs tend to be doing five things at least at once these days
This was the best series I've seen in the last few years. It's so good!
It's so great to have a mayor that actually supports congestion pricing!
Aquelas botas de luxo nem sΓ£o tΓ£o caros assim. O nypost tΓ‘ tipo obceΓ§ado com Mamdami ser "rico", mas comparado com os ricaΓ§os de NY de verdade, nΓ£o Γ© nada.
Different Andrew Hyatt, but Iβm glad you are enjoying it!
Some ridicule is I think fine, as are disagreements. What isn't so fine is ad hominens or people just being jerks. After all, we all have ideas that are not correct, right? And we are all people with normal human emotions. So it's important to have disagreements that are civil.
I think anything that makes the network an unpleasant place to be. If you have a network, you want your network to be a place people enjoy spending time on, and are able to have a reasonable discourse without getting flamed to hell. So for example, no spam, racism, sexism, abuse, general assholery.
I'll go even further - the solution to bad behavior is not blocking or even moderation, although those are necessary, just as the solution to spam calls is not number blocking at the user or carrier level. Instead, the network should make it hard to engage in the bad behavior in the first place.
No one likes to hear it, but algorithmic feeds are just better than following feeds, and the reason you mention explains part of why that is.
Threads and similar social networks have largely solved this problem but they are optimizing for engagement, which is horrible.
Thatβs super helpful. Iβll give it another shot, thanks!
I tried this and it was interesting but odd. I feel like I'm missing the vision behind this. comint is pretty well suited to the kind of interaction that these tools have. Also, most of the time it seemed to get stuck at the end of the turn (I was using yesterday's version of everything) .
I was fairly certain that Hochul wouldn't do the right thing here, but I was wrong. She's capable of doing something that could cost her votes, but is clearly correct. Props to her!
Random question: what's the recommended way to start a session with a request and have it be processed by a particular model? I can set the model, but only after the session is created. So I usually just ask the model to give an "OK", then set the model, then ask my question, which is awkward.
Everything the experts and people knowledgeable about congestion pricing said would happen has happened, and nothing the ignorant doomers whined about for years on X and in community meetings has happened.
bsky.app/profile/fund...
Great to hear. I really like the project, thanks for so much work on it!
My talk at #EmacsConf on different ways of using LLMs in Emacs and the future of editing: www.youtube.com/watch?v=U3kb....
Also, I apologize for the title card which makes me look manic.
I think current results show that it *is* a foundation for actual intelligence. It might be that there's some fundamental limitation to this technique, but it's extremely effective even at solving difficult problems. I don't think Gary Marcus is correct here.
When Wales lowered its default speed limit from 30 mph to 20, it created a perfect natural experiment. England, a nearby jurisdiction did not.
π You can see road deaths declined by 25% in Wales after the policy was implemented. Via @20splenty.bsky.social www.20splenty.org/gb2024stats
School Rumble dub was fantastic as well!
Thanks for posting this, I had no idea there was a movie made from Hard To Be A God, or that it was good!
I agree with you, but the problem wasn't Mayor Adams but the city council.
Good reaction time for sure, but the driving skill isn't so great, since he seems to be going way too fast for that kind of road, forcing him to take drastic evasive maneuvers.