digdugdirk

I would be careful about this one. While the overall impact (in the global/national aggregate sense) may not be massive, the impact to individual communities nearby these new hyperscale datacenters is far more impactful than most people on this site might think.

Look at the grok datacenter in Memphis for one example. The "move fast and break things" mentality in this arena isn't about code anymore, it's being applied to communities.

Again, we're in agreement, and I yearn for a world with massive buildout of scalable nuclear power.

But what is this "renewable lobby", and how are they doing funding-wise against the various extremely well funded and deeply entrenched fossil fuel industry lobbying groups (and companies!) that have been pouring money into UK politics ever since they stopped setting UK foreign policy directly and overtly?

All I ask is that people take a step back and look at the whole picture, and then question whether their arguments benefit an industry that is responsible for so much damage and destruction (environmental, economic, human health, societal, political, etc.) or if their arguments benefit individuals and society as a whole.

Nuclear power built out in such a way as to achieve the original "too cheap to meter" goal would be a dream. But don't let perfect be the enemy of good and all that.

It's important to look at the whole picture though - most of the issues you raised are political issues, either brought about by NIMBYism or a lack of infrastructure investment over the past few decades. Without this investment (which, yes, is more expensive than it would be if it weren't also providing additional benefits tailored for renewables) the grid would continue to deteriorate. So clearly there is some amount of this cost that would need to be invested regardless.

There's also subsidies for fossil fuels to consider [0]. I don't hold these figures as gospel, but there's inarguably a massive amount of money going to propping up the (wildly profitable and hugely destructive) industry that's causing most of your raised issues in the first place - either through reduced maintenance and infrastructure investment (gotta get those shareholder returns) or lobbying/public influence campaigns.

To be clear - I absolutely agree with most of your complaints. I just see them as issues caused/exacerbated by entrenched political players, and I think the benefits to our society of getting off our fossil fuel addiction are worth the costs of modernizing our infrastructure for the long haul.

[0] - https://www.theguardian.com/environment/2023/mar/09/fossil-f...

This seems tailored to the Claude web/chat interface. Does anyone have any experience or systems specific to Claude code?

I've been using Opencode alongside Claude, trying to utilize Opencode for as much easy/rote functionality as possible so I don't blow through my Claude context, but it is a pain in the rear. I'm sure someone on here has solved this for themselves, and I'd love to hear what people are doing in the "token efficiency" realm.

Can you explain this? I haven't heard of JEPA, and from a quick search it seems to be vision/robotics based?

Where are you getting this opinion from? Because the vast number of matriarchal hunter-gatherer societies throughout history would disagree with you. Those "woke" policies existed prior to the agricultural and industrial revolutions and were stamped out by the "mentally developed" societal and economic systems that we invented along the way.

I'm not sure where you're getting your "fundamental truths" from, but as the fields of sociology and economics don't actually have anything of the sort it'd be worthwhile to expand your reading list and adding some history in there for good measure.

Interesting. I tend to imagine the exact opposite - it seems as though this is the big players realizing there's not much left from training better models, so they're focusing on scaffolding to get their improvements. If that's the case, why wouldn't a company completely devoted to the scaffolding not do a better job at that goal? And more importantly, why would I lock myself into their system?

It's inherently limited by its geometry kernel. Most "real" CAD suites use something like parasolid, usually with a bunch of extras slapped on top. Making a new one from scratch is a massive undertaking, but I'll remain forever hopeful that we get a new, modern, open-source kernel one of these days...

Commenting here in case you or someone else remembers what this is. I'm always on the lookout for practice resources I can recommend to CAD beginners.

Honestly? I've been playing with using LLMs specifically for that reason. I'm far more likely to make prototypes that I specifically intend to throw away during the development process.

I try out ideas that are intended to explore some small aspect of a concept, and just ask the LLM to generate the rest of whatever scaffold is needed to verify the part that I'm interested in. Or use an LLM to generate just a roughest MVP prototype you could imagine, and start using it immediately to calibrate my initial intuition about the problem space. Eventually you get to the point where you've tried out your top 3-5 ideas for each different corner of your codebase, and you can really nail down your spec and then its off to the races building your "real" version.

I have a mechanical engineering background, so I'm quite used to the concept of destructive validation testing. As soon as I made that connection while exploring a new idea via claude code, it all started feeling much more natural. Now my coding process is far more similar to my old physical product design process than I'd ever imagined it could be.

Hmm... Definitely seeing my weekly usage tick up during the off peak timeslot, so like most things relating to our current AI ecosystem... be careful, your experience may be different than what is claimed.

"Does bonus usage count against my weekly usage limit?

No. The additional usage you get during off-peak hours doesn’t count toward any weekly usage limits on your plan."

Super cool! Do you do any analysis or have any tools that help you identify these circuits? I came across this [1] recently, and wanted to try to identify specifically strong "circuits" in what seems to be a similar way to what you did.

[1] https://weightwatcher.ai/

Honestly just wanted to thank you for this write up on the consumer side of tape storage. It doesn't look like my archival needs are at this level just yet, but this is an amazing overview and starting point if/when I get to that point. Thanks again!

What is the best consumer friendly long-term storage medium? Are we still better off with high capacity dvd/Blu ray discs?

Systems are complicated. Given there are numerous predicted outcomes (it's not just about the actual measured sea-level rise, after all) and many of those predictions are coming to pass far earlier than hoped, it might be worth having an open mind to the fact that sometimes people who devote their lives to studying something might be worth listening to.

There is a substantial difference between the standard lobbying and greasing the legislative wheels, and what's going on with this current administration.

Even if companies were pretending to play by the rules before, at least they had some need to put in the effort to pretend. When a society can see belligerent ostentatious corruption going on as the norm, nothing good can follow.

Very interesting! I hope it gets upstreamed soon, there's a ton of potential for "mental overhead" simplification in the nix ecosystem, this seems like it could be a huge help for that.

I'd imagine at some point the rig tolerances/vibrations/newly settled dust specks from snapshot to snapshot would completely negate any benefits you'd get from that level of detail. The processing power to handle that resolution would be a huge (but potentially interesting...) problem as well.

Can you please explain a bit more about why it's a difficult photogrammetry challenge, or point me in the direction of resources so I can learn more about it myself? This is an exact project on my projects list, so I'd love to have a better grounding in the topic when I get around to diving in to it.

Edit: I'm more focused on getting a dimensionally accurate/stable model, vs an esthetically pleasing one, if that matters. The hope is to be able to scan a broken chair and be able to design a jig in CAD that I could then 3d print for holding a specific piece in place while everything goes back together.

They don't get as much visibility into your data, just the actual call to/from the api. There's so much more value to them in that, since you're basically running the reinforcement learning training for them.

Looks very useful and very cool! Just a heads up - your graph loads terribly on mobile (android + Firefox), it's just a skinny strip in a container at the top of the page.

Aha! You might be just the person to ask about something that's I've always been curious about - are there any other types of Braille mechanisms other than the "pin on a lever arm" concept? They seem so fragile and clunky, and I'm surprised there hasn't been anything revolutionary that's sprung out of the miniaturization over the past 3 decades or so.

I'd love to see this paired up with Pydantic for a lightweight pydantic based configuration "language". Similar to CUElang, but using pydantic to describe the configuration models themselves.

How does this compare to the sPy project [1]?

[1] - https://github.com/spylang/spy

I'd love to see a breakdown of the token consumption of inaccurate/errored/unused task branches for claude code and codex. It seems like a great revenue source for the model providers.

That looks really cool! I've been looking for a more thought-out approach to hooks on JJ, I'll dig into this. Do you have any other higher level architecture/overview documentation other than what is in that repo? It has a sense of "you should already know what this does" from the documentation as is.

Also, how do you like Radicle?

Agreed. I do think the metaphor still holds though.

A financial jackknifing of the AI industry seems to be one very plausible outcome as these promises/expectations of the AI companies starts meeting reality.

That's not how complex systems work though? You say that these tools feel "symmetric" for defenders to use, but having both sides use the same tools immediately puts the defenders at a disadvantage in the "asymmetric warfare" context.

The defensive side needs everything to go right, all the time. The offensive side only needs something to go wrong once.

The Martian Revolutions podcast series is really good! As a bonus, it borrows from a variety of historical events from the past, so it can be a fun "side quest" to start picking up on those threads as well.

I've been curious how this project will grow over time, it seems to have taken the lead as the first open source terminal agent framework/runner, and definitely seems to be growing faster than any organization would/could/should be able to manage.

It really seems like the main focus of the project should be in how to organize the work of the project, rather than on the specs/requirements/development of the codebase itself.

What are the general recommendations the team has been getting for how to manage the development velocity? And have you looked into various anarchist organizational principles?