I wonder if they'll continue to support and develop Cog [0], which I've found quite useful for local deployments of models (as well as to Replicate's servers).
The one time I’ve interacted with cog, predictably, it delivered none of its benefits compared to a vanilla pyproject.toml, and all of the downsides of being, ironically, not replicable.
Not at all snarky because I have no idea what the real answer is here, but although this sounds pretty good, it strikes me how in conflict this would be with most peoples' ideals if we tried any sort of system to scale this to larger populations (like a city).
Perhaps we want privacy-but-not-anonymity. Or perhaps society doesn't scale easily.
The real answer is to rebuild the high trust society by relentlessly removing the low trust elements, the same way the high trust society was created to begin with.
That isn't a real answer. There is no effective way to remove the low-trust elements who physically operate in jurisdictions with no effective law enforcement. Like what are we going to do, fire cruise missiles at random Internet cafes in Nigeria and Cambodia?
We are heading for a period of a not-fully-connected internet. Already people are cutting off China and Russia from their sites because their bot traffic is either overwhelming or malignant.
Those are not alternatives. VPNs and botnets already exist. We have minimal practical ability to control Internet traffic or transactions between most foreign countries. All it takes is a single compromised device in the USA for the scam traffic to get through.
Come on, be serious. No one is going to cut any transatlantic cables just to prevent scams. VPNs make a huge difference: as long as you can get a route out somewhere then you can use a VPN (possibly to a compromised host) to make your traffic appear to be coming from another source.
> And you'd better believe wherever they buried the lines they'd have objections and expensive consultations about the disruption and the HoUsE VaLuEs caused by trenching, drilling and service structures.
But those are temporary disruptions. Overground lines are permanent.
The reason utilities and the Grid prefers overground is: it's cheaper. It's not better. It's cheaper.
Something I've seen discussed very little is that Claude Code can be opened in a directory tree of any type of document you like (reports, spreadsheets, designs, papers, research, ...) and you can play around in all sorts of ways. Anthropic themselves hint at this by saying their whole organisation uses it, but the `Code` moniker is probably limiting adoption. They could release a generalised agent with a friendlier UI tomorrow and get much wider workplace adoption.
I have it master my music. I drop all the stems in a folder, tell it what I want, and off it goes to write a python script specifically for the album. It’s way better than doing it in the DAW, which usually takes me hours (or days in some cases). It can get it to 90% in minutes, only requiring some fine-tuning at the end.
Wow, could you expand on this? What kind of effects can you get out of it? I’m somewhat skeptical that this could even come close to a proper mastering chain, so I’d be extremely interested to learn more :)
Any effect you can imagine. It could probably write a DAW if you wanted it to, but a “one-off” script? Easy. I think the best thing is when I tell it something like “it sounds like there is clipping around the 1:03 mark” it will analyze it, find the sign flip in the processor chain, and apply the fix. It’s much faster at this than me.
Note that there needs to be open source libraries and toolings. It can’t do a Dolby Atmos master, for example. So you still need a DAW.
That's fascinating. I generally mix in-the-box, so my mixes are close to commercially-ready before mastering, but I've experimented with a few of the "one-click" mastering solutions and they just haven't been it for me (Ozone's presets, Landr, Distrokid.) I've currently been using Logic's transparent mode as a one-click master which has been slightly better, but this sounds really compelling. I generally just want 16-bit 48 KHz masters anyway, so no need for Atmos. I'll have to try this out. Thanks for sharing!
That’s how I use it. I’m not a developer, and using Claude Code with Git turned out to be more complicated than I wanted. Now I just give it access to a folder on my Mac, put my prompt and any associated files in that folder, and have it work there. It works fine for my needs.
I would like a friendlier interface than the terminal, though. It looks like the “Imagine with Claude” experiment they announced today is a step in that direction. I’m sure many other companies are working on similar products.
Over the weekend I had it extract and Analyse Little but Fierce, a simplified and kid friendly DnD 5e and extract markdown files that help me DMing for my kids. Then it Analyse No, thank you evil as I want to base the setting on it but with LBF rules. And then have the markdown turn into nice looking pdfs. Claude code is so much more than coding and it’s amazing.
Indeed. I’m having success using it as a tool for requirements querying. (When a sales person asks “does product A have feature X” I can just ask Claude because I’ve got all the requirements in markdown files.
What this shows to me, as someone who has committed some of the unholy crimes above, is that people want their system, however esoteric, to come naturally to them.
I think reading docs, understanding a new system which someone else has designed, and fitting one's brain into _their_ organisational structure is the hard part. Harder than designing one's own system. It's the reason many don't stick with an off-the-shelf app. Including Org mode.
> What this shows to me, as someone who has committed some of the unholy crimes above, is that people want their system, however esoteric, to come naturally to them
I think this is a vocal minority. Outside of internet comment sections, most everyone I know doesn’t care that much about their todo list software.
The most productive people I ever worked with all had really minimal productivity software. For one person it was a Google doc with nested lists. I know several people who preferred physical sticky notes or 3x5 note cards.
A lot of the people I’ve worked with who built elaborate productivity systems and custom software weren’t all that productive. They seemingly spent as much time doing productivity rituals and rearranging their productivity software stack as they did doing actual work. I count the really heavy Notion users in this category because I’ve recently been pulling my hair out dealing with a couple PMs who think “reorganizing Notion” and adding more rules for Notion is a good use of their time each week.
The most extreme example I remember was the eccentric coworker who was building an AI-powered productivity tool that was supposed to optimize his todo lists and schedule them according to his daily rhythms. He spent so much time working on it that our manager had to remind him daily to stay on track with his real work. He was obsessed with “productivity tooling” but the productivity was secondary.
I've got to tell you that I don't think people like gwern.net and andymatuschak.org are procrastinating. A lot of these people are very productive in public. Walking through their zettel-sites is like walking through their minds. The best thing zettel-people have done has been to assume that everything that you're writing that was clearly inspired by something else should reference and direct link that thing. I really think we should caching and redistributing our references too, but the law...
The stuff they, and many smart people like them, are putting in their public notes are sometimes becoming the authoritative bibliographies of little specialized subjects. Their notes get referenced in journal articles.
edit: also, as far as I know, the goal of having a Zettel system is to be as lazy as possible. To have all your notes extremely networked so you can find them pretty quick, you can get surprised about what's in them because old stuff gets surfaced, and you can always find a place to add the thought you're having or the notes you're taking. You save all of that time digging though stuff and filing stuff and losing stuff, which you can use to take a walk in the park or something. To accidentally write a few books just because you had all your notes about some goofy subject you're obsessed with in one place and one day you're like "that must be about a quarter million words."
edit2: also, also, this conversation is very quaint. 10-20 years from now we will all have zettelkasten and we will never look at them at all because we will use AI to interface with them. I'm sure thousands are already in that world, I'm certainly working on getting there.
I think it's actually a selection bias. Who is more likely to spend a lot of time on productivity systems -- a person who is on top of their obligations or a person who is drowning in them. A naturally organized person can do with simple txt, they are already doing okay. A chaotic person can build whatever complex process they wish, they will still fail.
That’s been my personal experience. Spend plenty of time looking at all kind of options to optimize my ir my teams workflow. Then just fallback on pen and paper or some very simple excel spreadsheet. Something thinking about being more productive makes you feel productive.
Sounds similar to playing video games: the rules are simple, so once you understand them, you can feel mighty and powerful simply by accomplishing banal tasks. Makes for a great dopamine rush.
Eh, I don't know. I wouldn't paint with such a broad brush here.
Regarding productivity/to-do systems: on one hand I agree -- I know a few people for whom it's clearly a form of procrastination and really just need to get on with it. On the other hand, I myself was one of those people, but in hindsight I just hadn't found the right system yet and had real, legitimate issues with the systems I had been trying. Turns out, YouTrack is damn-near perfect for me. I use it both for work and for my personal life and I really, really love it, even for basic to-do lists. The things I was missing from standard to-do lists was the concept of relationships ("depends on," "blocked by," etc) and the ability to schedule multiple projects together on a Gantt Chart. Put those two features together and what needs to be done when and in what order is pretty much inarguable, which is precisely what helps me stay productive, as looking at a huge list and feeling overwhelmed about what to do next -- especially if I'm trying to be efficient or strategic -- freezes me in my tracks.
Regarding second brains: I completely disagree that they're not useful. My Obsidian vault is genuinely one of the single most useful things I have ever done for myself. There's nothing fancy about it, I don't use most of the features, but having a massive vault full of notes is truly indispensable in knowledge work.
I spent several years trying to make a custom todo system and ended up back where I started using CalDAV and a basic todo app and calendar. Turns out I was always procrastinating because I didn't want to force myself to adapt to something simple.
I used to use caldav but then stopped because there was once a bug on the server side where I couldn't delete events. The main thing I also don't like is there's no encryption or privacy really from the provider unless you go with a more a less proprietary solution like from an encrypted email provider. The closest I saw was EteSync but it requires special apps, and can include special bugs :)
I also don't need my immediate todo list on a calendar. I organize my to-do list simply as "Immediate", "Future", "Distant future", and then put things under heading. Sometimes I add a due date if there is one.
I just had a few markdown documents in a "todo" folder, eg <work>.md <project1>.md etc. Recently I changed it to org-mode because that's a syntax designed for the purpose. https://nvim-orgmode.github.io/ works excellently.
Never been an emacs user in my life. I spent about 5 minutes perusing https://orgmode.org/org.html and was able to do the same.
Using standard CalDAV it should be possible to encrypt parts of it. And just have a local proxy that decodes/encodes. e.g. Having fields like this:
DESCRIPTION: not secret
DTSTART:with week
X-mydata:
ENCODING=Ascii85;FMTTYPE=application/octet-stream
<~6q($A;Fs\a8P`)B+EM+(Eb0>"6r[)a5uLZCGA2/4+D>\9EcVQ~>
and then just decode that and replace the X-mydata with the decoded data:
DESCRIPTION: secret
DTSTART: with hours
perhaps support for removing existing fields that is in the encrypted blob as well, if you do not need state in the encryption it should be good. I do not know how to create a good encryption protocol so I am sure there are lots of stupid ways to mess this up. I only need to encrypt the content of my description and time I never need to hide that I created a meeting when I had a meeting with a Spy from the NSA.
Most of my daily agenda is also stored in my employer's Google Workspace. CalDAV is the common protocol for exchanging and storing events. Most of my task entries don't have a due date, I just sort them by priority.
I understand the privacy concerns, I can host a nextcloud in a secure location if I need it.
Totally agree, but I have learned that exciting isn't necessarily better. I just got busy enough that I had to accept the boring option to stay afloat. Which was a certain kind of excitement in itself.
> Same. I dabbled in the second brain fad for a while, until I realized I already have a first brain.
The way you chose to describe keeping engineering notes as "second brain fad" is telling. It says you mindlessly tried to follow an organization scheme even though you felt no need to adopt an organization scheme.
In other words, you somehow hopped onto a solution searching for a problem you didn't had.
That's perfectly fine. Fads are defined by those who, like you, onboard onto something for all the wrong reasons and without spending any time thinking about what they are doing and why they are doing it.
Of course, those who end up searching for problems that fit a specific solution end up not finding it. That's the fad part.
In the meantime, engineering logs are indeed time tested and a tried-and-true technique. Those who use it to solve problems they have will naturally see their problems solved by them. That's why you see blog posts like this, and people commenting on how they scratched their itch.
There’s a now quite dated comment from Merlin Mann: "Joining a Facebook group about productivity is like buying a chair about jogging.”
It’s fuzzy - but my recollection was Mann was a fairly renown productivity influencer (although I guess we wouldn’t have called it that then), who had an apostasy about it all.
I think he had a blog called LifeHacker and/or 36 folders (I don't know if they were his or just a writer, but I remember following him back in the day).
It's wild to think that was almost 20 years ago when Getting Things Done was going through tech circles as the organization method dujour. (I also equate the same time period with learning Ruby on Rails.)
Merlin ran 43folders.com[1]. That domain name was a reference to GTD's tickler files[2].
Merlin did not have anything to do with Lifehacker.com. Gina Trapani[4] founded Lifehacker[3]. But, Lifehacker often covered and was inspired by Merlin's work.
The real takeaway from your story is that it's easy to stay on task when you're interested in the task. Your coworker just didn't care about his work. But if his work was creating a productivity tool then he'd probably love his work and be productive.
On managers side the equivalent is making fancy JIRA workflows with all the fancy fields so that everyone is informed. Makes people annoyed with extra work and that time could be spent just talking to people to understand what's actually happening.
Exactly. It takes enormous effort to get product and engineering teams to agree on how to use JIRA properly, because everyone has their own ideas around how and what to organize. It's exhausting.
OR to put is differently - everyone has their own needs they are trying to get done. Each one need individually is simple so it is easy to demand that need is added too. Until the whole system for each different need becomes so complex it collapses under its own weight and you move to a new tool. This cycle seems to repeat every 10-15 years at all companies. JIRA is the tool everyone talking about today, but there were many others in the past, and there will be a different one in a few years.
Generally the tool isn't the problem: NEVER put ticket numbers into long term storage as in a few years you won't be able to reference them. That is version control, design documents, and anything else that isn't the ticket system itself. You can talk about who is working on ticket 12345 and the problems they face, but if anything is going to be written down you need to summarize the ticket without a number.
So much of it is empty productivity, all prepping for the work but never actually doing it.
Like the old joke about the programmers spouse who died a virgin because every night all the programmer did was sit at the edge of the bed talking about how awesome it was going to be when they finally did it.
I’m very guilty of trying all sorts of productivity software as a form of procrastination. The best one did, in fact, turn out to be index cards and a pencil.
The phhsical copy served an important purpose: it forces you to admit you will never do something and so give up on is. until I die it is safe to assume I will eat 3 meals per day. (It won't be 100% because of sickness but close enough) thus if I'm out of some food I will need a todo list to replace it. However if I never finish the ukuele I've started it won't matter and it is reasonable for me to give up on it.
> The most productive people I ever worked with all had really minimal productivity software.
What about quality? Often, people are very productive, because they sacrifice quality for speed, especially the "annoying" longterm-values of products/decisions.
> They seemingly spent as much time doing productivity rituals and rearranging their productivity software stack as they did doing actual work
It's a different kind of productivity. Just not as valuable for the company.
The term that comes to mind, and one of my favorite concepts, is "progressive disclosure", which is a concept we really ought to be more mindful of.
One of the perks of just-a-text-file-with-a-bunch-of-addons is that it enables progressive disclosure - it takes no learning curve to just get in and use the tool on a basic level, but additional complexity (and power) can be introduced over time.
The problem with a purpose-built app is that there's a minimum level of new concepts to learn before the tool is even minimally useful, and that's a barrier to adoption.
A good example of this in action is something like Markdown. It's just text and will show up fine without you learning anything, but as you pick up more syntax it builds on top - and if you learn some markup syntax but not others, it doesn't prevent you from using the subset you know. There is a clear path to adding new knowledge and ability.
Right, instead of fomo over not using the extra features of utilizing the right flow - people tend to experience the want/need to incrementally increase complexity when using roll-your-own software
The whole point of org-mode is that it's so malleable, that you can extend it to be whatever you want it to be, much easier than writing your own, ad-hoc, bug-ridden reimplementation of org-mode.
This makes it so infuriating that the top comment on Todo systems is almost invariably "just org-mode lol". Same as remote editing "just TRAMP lol".
I am not going to completely change my editor and rebuild two decades of optimization just to use two Emacs tools.
On-topic: TickTick or Todoist with a slimmed-down "Getting Things Done" system works really well. Almost no learning curve, and you get to free up so much mental bandwidth vis a vis remembering things and prioritizing things. And you don't have to do hamfisted tricks to make a 'simple' .txt system work. Bliss.
> I am not going to completely change my editor and rebuild two decades of optimization just to use two Emacs tools.
Change your editor and rebuild two decades of optimisation in order to use Emacs, two Emacs tools, and also every other Emacs tool out there. Org Mode, TRAMP, Magit, gptel, eglot, flycheck, elfeed, ERC, Emms, EWW … there are a ton of reasons to use Emacs.
Or you can keep using less-capable systems and being annoyed when folks recommend that you upgrade.
Your argument highlights its own flaw; changing your editor opens up a world of tooling that's certainly adequate for most use cases you can throw at it, but it also requires either discarding or (worse) un-learning all of the tooling that you've learned for your current editor.
For example, I'm perfectly content to use nvim as my primary editor, and this was born out of having to develop for and administer literally tens of thousands of linux servers professionally. I have all the plug-ins and configuration necessary for productivity on my development machines, and when I'm on a remote system ad hoc editing a configuration it already has a built-in lightweight version of the editor I'm already used to.
If I switched to Emacs locally, I'd still have to maintain a working knowledge of vi and context switch when in a remote shell. Changing to Emacs would require more cognitive bandwidth when the whole purpose of "switching for org mode" is to reduce mental load.
> If I switched to Emacs locally, I'd still have to maintain a working knowledge of vi and context switch when in a remote shell.
Even ignoring the possibility of installing Emacs on remote systems, there are still alternatives:
1. You can run remote shells within Emacs, and edit files remotely using TRAMP. When you are editing a remote file, shell commands run from Emacs run on the remote system.
2. You could use Evil, the Emacs implementation of vim. Then you would use the same bindings everywhere.
3. I have been running Emacs locally for literal decades now, but I still remember and use vi frequently, both locally and remotely. It’s really not a problem.
I feel like there must be an editor version of the Blub Paradox.
> I am not going to completely change my editor and rebuild two decades of optimization just to use two Emacs tools.
Fair point on the surface, it's missing the key aspect of what Emacs actually is. Emacs is not just an editor - to a degree it's philosophy is that your computing environment should be malleable. Those two decades of optimization don't get thrown away; they get encoded directly into the system. Instead of learning to work around the limitations of separate tools, you're investing in a platform that can absorb and amplify all that accumulated knowledge. The question isn't whether you want to abandon your workflows, but whether you want to be limited by them forever.
The key insight is reframing it from "starting over" to "finally having a place where all that expertise can compound indefinitely."
I mean, I get it - not everyone wants their editor to be infinitely customizable; sometimes you just want something that works out of the box. Yet I do honestly think every programmer should give Emacs a serious try at some point in their career - not because they'll necessarily stick with it, but for the same reason why everyone should learn at least one OOP language and get introduced to an FP language. It expands your thinking about what's possible.
Even if you go back to your previous tools, you'll understand computing differently, having seen what a truly malleable environment looks like, the idea of a truly personal computing environment - one that grows with you rather than constraining you. Dismissing Emacs as "just another editor" misses what makes it fundamentally different.
I completely understand your impulse (been there myself), but I'd encourage you to keep an open mind about what Emacs actually offers. When you get a chance maybe explore the philosophy behind it. You might discover something unexpectedly rewarding.
Agreed on TRAMP. It's great and all, but not worth abandoning your toolong.
org-mode though... It's called Emacs' killer app for a reason. Even if I only used Emacs for org-mode it'd be worth it. And I don't even use the productivity features.
What I gather is people really like the blank whiteboard. There’s something about Notepad and Excel, the freedom, the linitlessness, of having a blank canvas and being able to do anything.
Todo software is too opinionated. It’s not flexible enough to allow you to break rules. You can’t move things around in a way that allows you to control visual white space between entries. Everything “is something” (a task, an event) vs just being (text.)
If the tool brings you (you, as in: the one building it) joy and gives you a chance to let your mind float, then yes, it is useful. Just not for others. And that's fine.
Exactly. Most people wish they could customize their Todo app or system to their specific preference or need, but have no way of making it happen. Devs can, so they do.
What's interesting is AI is going to change this. Entering a prompt for an app that has all the features you want is already pretty trivial, and will only get better.
Which is why everyone likes spreadsheets. You learn a few formula and styling rules, and you’ve got a hammer for every nail that’s been bothering you (if it was actually a screw…shrugs).
Yep, at the risk of repeating what you said: I think this is why so many project management / todo apps exist with their own flavor of the very basics. It's a reflection of those wishes that feel natural to us individually, and it just so happens many of these apps mesh well with our model of thinking and organizing.
This is also why it's so difficult to get teams on the same page about project management in their respective workplaces.
I think in part because larger systems aren't typically custom made to the user's exact workflow (especially because users don't typically have one single workflow anyway!). So not only do I have to get into someone else's mind, but it feels ill-fitted to my own mind. Thus, it's also more inefficient.
Note taking and task management are two things which everyone has a slightly different style and need. There is no one size fits all and in a group someone will always find some aspect lacking.
The same has happened in the United Kingdom wrt deer and predators including wolves (which are locally extinct) -- a blunt instrument because we can think of no other way of protecting livestock. The result is we "have to" regularly cull thousands of animals instead of letting an ecosystem manage itself.
What would that entail? The whole of the UK has been a human managed ecosystem for centuries. Deforestation was completed about 400 years ago and the larges stand of contiguous trees is under 300 sq. mi. So many of the species that would have made up the old ecosystem are gone.
Here in Germany wolves return to many parts of the country. And there is lots of resistance. There are regular sob-stories how wolves hunted poor cuddly niece lambs and how their owners are now scarred for life and will quit their jobs.
There is financial compensation and also guidance on how to build secure fences, but the big bad wolv is scary...
If the circle of life scars somebody out of farming, that's probably for the better. Farming is hard work and any time you have livestock, you have to make hard decisions about managing them, including how to protect them from predation.
Preventable via multiple methods, you've been downvoted presumably for seeming insensitivity, but it seems a valid question. Water is plentiful, it's a rich area. This stuff can be prevented. Why wasn't it?
And a surge in road deaths.