Admittedly, I’ve gone through waves of feeling scared for the future of software engineering. I manage many people so I care about them and their futures. But the more I use AI in writing software, not just vibe coding, the more I align with this perspective. And the way I see it, when AGI does come it’s going to affect a whole lot more than the software engineering job market. That’s global societal-level impact type stuff. We’d have to reckon with if work (meaning a 9 to 5) is truly a requirement to thrive in life. Spoiler, it isn’t.
> Using existing CLI directly: No context wasted on tool definitions
Can someone explain this to me? I've seen claude code try to run a not-well-known package and it basically shot in the dark a command, noticed that failed, then ran the help command for the cli tool to get a list of commands and what they do.
How is that different than passing the tools with an MCP? Like how are we saving context?
This is also my point of confusion. People in the comments seem to be saying that MCP is necessary due to discoverability, but I fundamentally fail to understand how a protocol can make interfaces discoverable to an LLM in a way that wouldn't also be achieved by making traditional interfaces more discoverable to a human. These things mimic human behaviour after all.
Surely people aren't saying we haven't solved API discoverability by now and need new tech for it.
The usual problem is companies write an MCP server with 50 different tools, and each one has a schema, description, etc. Say each tool is 150 tokens, that's 150 * 50, or 7500 tokens, dumped into the beginning of every session. Compared to a text file that gets loaded on demand with command-line tool examples, so you still get close to the same amount of context, but you can control what tool definitions you pull in.
The other thing is the agent gets the entire MCP API response dumped into context as a tool response in JSON, which can be a lot. Compare that to shell commands where agents often `head` or `tail` or `grep` the response (which I kinda hate, but it does save tokens).
It also depends on whether the agent loads them on-demand or not (most modern agents do), and whether your MCP has a ton of tools or not. If your MCP only has 2 tools, and the responses aren't big, it's really not that much context.
The other thing that doesn't get talked about is the non-determinism of shell one-liners. There is a lot more non-determinism in shell tool calls; the AI can mess up commands, options, arguments. It can incorrectly filter output, miss output, miss return status, which results in re-running calls, polluting context, making results worse. Compare that to MCP calls which are more likely to succeed because they have a schema, well-defined errors, etc. Do you want less token use or more reliable results?
The thing is, you don't have to pick a side. I personally use both MCPs and CLIs at different times in different ways. Often I'll have the AI write a small script to do many calls (sometimes with tools, sometimes with libraries) which saves tokens, allows me to review, and is more deterministic.
Not all workers will fare equally. As an illustration, Reuters cited a union source estimating that someone in the memory chip unit earning an 80-million-won base salary could take home roughly 626 million won in total bonuses this year. By comparison, workers at SK Hynix stand to collect upward of 700 million won should their employer post annual profit of 250 trillion won, Reuters calculated. Unlike at Samsung, SK Hynix employees are not limited to stock payouts and may instead opt for cash, Reuters reported.
I love accessibility, I just want to preface what I’m about to say with that.
I found this site hard to read. I’m reading on my phone btw.
The text is too big for me and the line height (space between lines really) isn’t right, it’s too spaced out. Can I read it? Absolutely, I just can’t read it as fast as I normally would. It’s like when my mom hands me her phone and the text is so large I can barely operate it for a while, then I eventually get used to it to a certain extent.
What’s funny is this itself is an accessibility issue in the opposite direction of most accessibility issues. Just goes to show users should really be able to have their own text preferences reflected on the web.
I agree, I found it hard and frustrating to read on my (small) phone because the text is just too big. I usually skim long articles to some extent to focus on reading the parts I'm most interested in, but this format makes that impossible. I can't skim anything because barely a sentence is on my screen at one time.
fwiw I got one of the first products listed on amazon when you search mycorrhizal fungi and I'm seeing the same effects stated by the grandparent comment
Given that Plex just bumped their lifetime subscription price to $750, I can no longer recommend them. They are clearly more interested in becoming another streaming service, and are I think trying to push out their core users who probably make them very little money.
Interesting, their price bump announcement actually just went and made me upgrade to lifetime (at $250 while I could) instead of write them off completely.
Netflix will never allow you to pay a one time fee for life, neither will any other streaming service on the planet.
Meanwhile, plex is a company that has employees. If I like plex, use it heavily, and want to support them I can do so with money. There are alternatives that are completely free, but I don’t like them as much and the minimal cost for plex is totally worth the value for me.
Goes to show how inoculated I am against banner ads I suppose.
Either way part of me feels like it’s for the best. One off payment for lifetime membership of an app that has continual development isn’t a great business model.
the only things i care about is some essy enough to use upload process, basic serving, then that theres some smart enough local caching on whatever device im using
I mean.. did he turn anti-Elon or did society turn anti-Elon when Elon started doing insane shit? Reminder: Elon bribed his way into leading a self-formed and now essentially defunct government organization to "save costs" but failed to do so in any meaningful way. He did that like a year ago. Meanwhile, the dude made BILLIONS off of government tax breaks, and still leaches off the government with his other companies.
Frank being against Elon speaks less about Frank and more about Elon in my eyes.
> The real danger was not allowing smoking because if you ban smoking, people will smoke no matter if it's banned.
This same concept is why full prohibition never works. People who want to do something will find a way and it often comes at the cost of being more harmful to society than if they were allowed to do it in a controlled environment.
That's always what I've said too, so I'm now proposing to put all the "want to be murderers" together with folks who want suicide assistance. Make one move, get two results or whatever they say.
Yep, many older people right now don’t have a smart phone and never will.
As long as some younger people stay that course we should be fine. Hopefully we’ll see an increase of dumb phone adoption in a growing cohort younger adults. But the FUD spread in threads like this actually spreads misinformation and makes that less likely to happen
And those older people frequently have to ask someone in their life who has a smartphone to assist them. They still need a smartphone to manage modern life they just don’t happen to own the phone they need.
> Perhaps rather it is management that is wiping out those jobs
As a manager I’d direct you to the actual decision makers for things like this, company leadership teams. They’d blame the market, yet most of the big tech companies laying off or freezing hiring are doing quite well financially so it makes you wonder.
reply