For me GPT always seems to get stuck in a particular state where it responds with a single sentence per paragraph, short sentences, and becomes weirdly philosophical. This eventually happens in every session. I wish I knew what triggers it because it's annoying and completely reduces its usefulness.
Usually a session is delivered as context, up to the token limit, for inference to be performed on. Are you keeping each session to one subject? Have you made personalizations? Do you add lots of data?
It would be interesting if you posted a couple of sessions to see what 'philosophical' things it's arriving at and what proceeds it.
I feel the same way. Given it's AI-written, looking at the code isn't even worth it to me. I would rather read a blog post about how they develop it day to day.
That's not it. Flame wars are as old as the internet. The quality of discourse has plummeted largely due to these factors: 1) democratization of access to an audience and 2) engagement maximization algorithms. Anyone with a hot take can post it, get people angry fast, at which point the engagement maximization algorithm picks it up and carries it far and wide.
Focusing on "copying" seems like missing the forest for the trees. There's the copyright angle, but copyright laws are unnatural obstacles designed to give the original author some control over what happens after publishing. They're not fundamental, we made the laws.
What is fundamental is this: every artist starts out by copying the works of others. It's how you learn.
And in that framing, once you publish your derived work, there is only one question that arises - if you don't credit the original author but sign your own name, you're fundamentally misleading your audience. Your audience implicitly assumes you made the thing. Maybe you made 95% of it, but if you don't give due credit, it looks bad once your audience discovers that.
On more than one occasion my perception of an artist has shifted once I discovered the "brilliant work" they created was actually a remake of somebody else's brilliant work. It's a feeling of being misled. It's never a feeling of "wow, this guy is a total hack and has no ability of their own".
They wouldn't copy each other for copyright infringement as much as it was a mark of respect. They carried each other's arts as an evolution and respect towards each other rather than copying; all bringing a small twist on what was before.
> On more than one occasion my perception of an artist has shifted once I discovered the "brilliant work" they created was actually a remake of somebody else's brilliant work. It's a feeling of being misled.
The spirit of the famous - cited in the TFA as well - quote "great artists steal" is exactly that. If you don't know that the inspiration came from somewhere and believe that what an artist did was created in a vacuum, you will certainly think much higher of said artist.
This sounds so insane to me. If I own land and grow a tree on it, the tree and its fruits are private property forever (mine until I die, then inherited by my children, then their children, or sold, transferred, etc ad nauseam). At no point does the tree become "public", that would be utter nonsense. It is property. Why should my ideas then be anything different? They come from my head. I own myself, including my head, thus I should own the fruits of my head like I own the fruits of my tree and they should remain property forever. The fact that copyright expires is one of the great tragedies of modern life, though at least I can take solace in knowing I own my ideas until I die.
Copyright law exists exactly because it is universally accepted that ideas are not property: Copying an idea or expression of it does not deprive you of your ideas.
The entire notion of "intellectual property" is the creation of an artificial monopoly rooted in very distinct and separate goals from physical property that requires separate laws if you want to restrict copying or exploitation, because property law explicitly does not cover them.
Most copyright laws are also justified implicitly or directly in the legal texts allowing them as creating an incentive for the advancement of the arts and sciences - a temporary monopoly right granted by the state as a deviation from perceived "natural right" - on the belief that granting that right creates more benefits for the public than not having them, by encouraging the creation of more works.
And no copyright law protects your ideas. They protect the specific expression of them. Patents - which do protect ideas - are by design far more restricted and limited exactly because they are far more invasive in depriving the public of use of the very idea for the duration.
Did you spend your entire life in isolation from the rest of human society? Because if not, then you have been influenced throughout your life in a multitude of subtle and not-so-subtle ways by the works of others. In what way, then, are the fruits of your head entirely yours? We're all standing on the shoulders of giants.
Unlike the tree, nobody can take your idea away from you. You retain possession of your idea even if somebody copies it. It sounds insane to me to think you should get permanent control over other people's communication just because you had the idea before them.
Perhaps, but that is your tree, if someone takes a cutting from your tree and grows it into their own tree you shouldn't own that tree, your tree is still there.
Eventually you get to the point where someone asks why the tree is theirs and they say it's because someone in history planted it, they were a relative, so it is mine now. It is hard to assert a moral justification for long term hereditary ownership without inviting investigations on how it was those ancestors came to have the resources that caused the ownership to begin.
It's not a "should" -- ideas simply are fundamentally different from physical materials, and the norms we use to deal with the inherent qualities of one don't automatically translate over to the other without a suitable rationale.
Physical materials qualify as property because they are economically rival: one party possessing and using them inherently excludes others, meaning that competing claims to the same thing must be resolved by one party surrendering their claims to the other. There's no agree-to-disagree mechanism available, so we need a way to resolve disputes in favor of one party or another.
There is no clear application of this to non-rival intangibles: there is no conflict between two people using similar ideas independently of each other in the first place. Someone copying your idea isn't analogous to them picking fruit off of your tree, it's analogous to them learning from what you're doing, and then going off and planting their own tree on their own land.
Modern "intellectual property" is a contrivance by people desiring to artificially incentivize certain categories of activity by attempting to replicate one of the downstream effects of the inherent exclusivity of goods, namely commercial markets. So you wind up with a positive-law intervention to create artificial scarcity in order to produce similar second-order consequences to what comes about when scarcity exists naturally.
That's why property rights have been recognized in all civilizations in human history -- and are likely a prerequisite for organized civilization to exist in the first place -- whereas copyright laws in their modern form date to the 18th century.
In fact, artificial "intellectual property" conflicts with natural property rights, in that in claiming a universal monopoly on arranging any bits of matter into particular patterns, you are actually claiming the right to stop people from using their own actual property as they please.
Such a weird take. What are the similarities between your fantasies and land that to you make the philosophical convictions involved in private property laws applicable to those fantasies? Why isn't it good enough for you to fantasise about land and a tree, and why doesn't the answer to this undermine your reasoning?
Personally I'm not convinced by the arguments for private property, which makes your comparison even weirder than you likely intended.
This is entirely based on the "agent skills" system. LLM agent only sees the one-line skill description in its context and "lazy loads" the rest of the skill file on demand.
The lazy loading approach is smart. We've been publishing agent skills too and the context budget is a real constraint; six skills with reference docs would blow past 30k tokens if loaded eagerly.
Filtering at load time based on what the agent actually needs makes a huge difference. Curious if the orchestrator/executor split causes issues with state handoff between the two context forks.
We're using Augment Code heavily on a "full rewrite of legacy CRM with 30 years of business rules/data" Laravel project with a team size of 4. Augment kind of became impossible to avoid once we realized the new guy is outpacing the rest of us while posessing almost no knowledge of code and working fully in the business requirements domain, extracting requirements from the customer and passing them to AI, which was encoding them in tests and implementing them in code.
I'm using `auggie` which is their CLI-based agentic tool. (They also have a VS Code integration - that became too slow and hung often the more I used it.) I don't use any prompting tricks, I just kind of steer the agent to the desired outcome by chatting to it, and switch models as needed (Sonnet 4.6 for speed and execution, GPT 5.1 for comprehension and planning).
My favorite recent interaction with Augment was to have one session write a small API and its specification within the old codebase, then have another session implement the API client entirely from the specification. As I discovered edge cases I had the first agent document them in the spec and the second agent read the updated spec and adjust the implementation. That worked much, much better than the usual ad hoc back and forth directly between me and one agent and also created a concise specification that can be tracked in the repo as documentation for humans and context for future agentic work.
Far better results compared to GPT 5.4 and Opus 4.6. Not great for execution due to speed but has consistently had better comprehension of the codebase. Maybe it's a case of "holding it wrong" regarding the other models but that's been my experience.
Not in disagreement, but based on how casually the frontpage throws around terms like "X25519 keypair", it is obvious that mass adoption and ease of use aren't among the goals of this project. Looks more like an exploration of a concept - can a social network viably exist without any middlemen in between.
> The ~foo as backup convention is not part of any standard.
Emacs does foo~ by default, not ~foo.
In either case, you're not really supposed to edit files in sites-enabled. That directory is expected to contain symlinks to files in sites-available. I'm not going to say with any certainty that one of the reasons for this indeed is that the pattern (which was used by apache as well - and perhaps other things before it) protects against accidentally reading backup files, but it's not impossible.
So there's definitely a case of holding it wrong if you end up with backup files in that directory.
I liked doing symlinks so the site configuration is with the rest of the site, but that was before containers when it was common to host a bunch of sites on one instance apache or nginx.
> The ~foo as backup convention is not part of any standard.
> [...]
> It's the second thing I fix in either Vim or Emacs: Put backup files in a central location. (The first is proper indentation/spacing rules.)
Perhaps not a standard, but you yourself admit it's the default behavior.
Though I agree that the simple mechanism acts ... er,... simply, shouldn't it be at the very least aware of the default behavior of common editors?
But it DID contribute to the conversation ... it corrected an error/misunderstanding. If I didn't think I was contributing, I wouldn't have commented. You seem to have a different ethic.
I won't comment further, as we are now off topic (thanks to you).
Cool tip - never knew this. I always figured piping to `tee` is a must in order to view-and-save command output at the same time. Turns out I can do "command >&1 >file.txt" instead!
The biggest roadblock to using agents to maximum effectiveness like this is the chat interface. It's convenience as detriment and convenience as distraction. I've found myself repeatedly giving into that convenience only to realize that I have wasted an hour and need to start over because the agent is just obliviously circling the solution that I thought was fully obvious from the context I gave it. Clearly these tools are exceptional at transforming inputs into outputs and, counterintuitively, not as exceptional when the inputs are constantly interleaved with the outputs like they are in chat mode.
> I thought was fully obvious from the context I gave it.
Lot of people think they have given the right instructions but in most cases people miss some crucial points and that leads the model in the wrong direction, then the same people complain AI is not good.
reply