I don't understand how some of y'all use these things. I get garbage unless I give them very specific concrete tasks with as much context as possible. Anything that takes more than 30 min is usually a waste because the scope was too large.
Different people just have different concepts of what's garbage and what's not.
There seems to be some kind of AI hysteria going on, with people becoming so enamoured with the AI that they accept anything it produces as if it's some gift from the gods, while others just reject it prima-facie.
For example, the worst design I have seen recently was from a designer who pivoted into "vibe coding influencer". The worst code is from developers who were heavily into Clean Code a couple years ago and now half their PRs is unused dead code.
I had good experiences doing multi-hour refactoring/housekeeping tasks that basically consisted of applying the same steps and rules n times.
Worth noting, a significant chunk of those runs involved the agent waiting for the compiler, linters, type checks, and test suites, as well as updating journals.
It’s not the agent sputtering out code for eight hours straight.
And naturally I spend more time on manual verification in the end as much less of it is happening during the coding process.
In this case, handling all the edge cases and variants, and testing a codemod, would have taken significantly more of my time, which costs quite a bit more than the LLM.
Obviously, a deterministic tool is preferable in general, but it is not always worth bothering with for a one off task.
I usually make the llms do that part for me. Instead of asking the llm to refactor, ask it to write the codemod script that'll refactor, have it test that script, and even have it run it on its own. It's definitely faster and less error prone that way for me.
Neither is your time writing that prompt. When people are talking about elaborate prompts, with a lot of detailed instructions, guardrails etc. I'm kind of assuming it takes time.
I do this too, with a document written for this purpose.
> ... a significant chunk of those runs involved the agent waiting for the compiler, linters, type checks, and test suites, as well as updating journals.
That is a good point. I'm mostly using C, which seemingly compiles in O(1) time, so I could imagine a large C++ or Rust codebase taking much longer to iterate simply due to compilation times.
Clear evaluation function for an objective metric if they are making progress or regressing.
Evaluation function is computed, not llmed.
Ontology of potential actions clearly specified.
Accurate inventory of the current status qou.
Clear enumeration of options from status quo towards the winner's circle.
Waypoint objectives with similarly concrete evaluations of pass/fail, or on target off target.
It's the same thing when leading a large organization to actually hit a goal. There's randomness every turn away from your mind, so the more constrained the options, the more likely you are to hit the target. The consequence is if you're wrong about the plan then with people you're fucked. Morale will plummet. With AIs, they are so nerfed emotionally now, you clear context and start again.
I did enjoy Sonnet 4 when they would swear randomly and become sullen or wax desperately. That would at least cause pushback against a bad plan.
This is my fucking life at work right now. I look forward to the weekends. I've never been truly inconvenienced by shitty devs because they're often too lazy to really spam me with bad code, but now they are all free to do so. I spent so much time today writing guardrail markdown files when these people SHOULD HAVE BEEN ABLE TO REVIEW THE OUTPUT AND KNOW THAT IT WAS BAD.
It truly is the age of the 90 IQ software engineer. They've never had it better.
As if meetings weren't bad enough already, I now have to sit through an informal introduction to the model of the week and its personality characteristics and how quickly it burnt through one subscription's token allotment or whatever and the latest tweaks on the magic markdown files. Luckily I've only had a couple changes sent my way so far, which weren't much different than just getting a bug report to debug and fix myself. I will need to get into risky options gambling or something so I can go start my farm early, if it keeps going this way. Even supposing it all works correctly, I don't see how it is in any way enjoyable, satisfying, or fulfilling.
If you're giving it 8 hours of stuff to create with a template (e.g. slop forking) that's not a big deal. Letting it run for 8 hours to debug a weird failure also tends to work out.
Historically a $1T market cap with a PE of 20.0 would be achievable with a $50B/yr profit. That seems easily achievable eventually for SpaceX, as it has actual hardware and services and IP.
> Historically a $1T market cap with a PE of 20.0 would be achievable with a $50B/yr profit. That seems easily achievable eventually for SpaceX, as it has actual hardware and services and IP.
It seems crazy to me to make a comparison between a company being valued on it's current profit and then to say it's reasonable for another company to have the same market cap because it could eventually have the same profit.
Your sentence was the direct response to the question:
> Yeah, but is SpaceX actually worth $1T
The question and context of your response was about SpaceX's present value. Given the context, I think my response was absolutely a fair reading of your response.
Markets are forward looking, pricing in future cash flows. No one should be thinking about SpaceX's present value when buying their stock. If you do that you'll always be too late to the party.
That said, I don't agree on the valuation, but I can see where some people could be coming from.
It does have real hardware, but it’s not in wild growth areas. They’re making their most consistent money from Starlink, which is a solid product but has growth limited by competitors from conventional ISPs with far-superior fiber networks, and the space launch business is similarly not the kind of thing where you get Google/Facebook-level growth curves. That’s not a slight, it’s just different industries: advertising companies can grow rapidly because scaling customers is so much easier than launching cargo into orbit.
The wildcard there is AI, and that seems especially dangerous to project long-term revenue from their current performance: xAI is barely in the market except renting capacity to Anthropic, so you’re gambling that they’ll continue to pay $1¼B/month for what is largely a commodity offering. Even if you’re bullish on Anthropic, that doesn’t mean xAI gets part of their profits, and given the way they blindsided the local authorities there’s a substantially greater than zero chance that they’ll get a major setback if the neighbors win their lawsuits. That doesn’t mean they’re doomed, but anyone estimating their future performance has to factor in some real risks.
Yet, I, a relative peasant financially has been hit by 3 different brokers that I'm eligible to participate in the offering. I would hazard a guess they are not getting the uptake in institutional money they were hoping for.
> I would hazard a guess they are not getting the uptake in institutional money they were hoping for.
I would say exactly the opposite. They (really, Elon) want more retail uptake. This follows a similar strategy that Tesla used. Also, retail ownership is much less likely to be disruptive during shareholder meetings (proposals, objections, etc.).
As we enter this era of far more qualified candidates than jobs, HR will die eventually the equivalent of index funds is for hiring. Just as most money managers didn't beat the market and lost out to Bogle's low-cost index funds, people will figure out that HR doesn't do any better than any other random criteria for hiring and firing employees, since most of the applicants for most jobs will be able to do the job sufficiently well. Probably the answer is some sort of AI, but I bet you could do just as well rolling dice.
If most of us are honestly with ourselves, we'd realize the marginal return on difficult hiring decisions is extremely small.
As for the CYA aspect of HR, an AI can definitely do that cheaper and more callously.
That’s a very simplified view of HR. Human Resources does a lot of things. During the cash rich days a tech company HR department might end up doing a bunch of nonsense[1] but the fundamental value of the department is there and will continue.
HR is rarely involved in hiring decisions, that’s the responsibility of the hiring manager which is typically the new hire’s manager. At a very big company you might have HR screening applicants but that’s to save time for hiring managers.
[1] just as engineering ends up doing a bunch of nonsense when the money is flowing.
For me a TUI constrains the style of presentation to something reasonably good. In an era of websites with tiny light gray font on dark gray background, where scrolling is really a javascript page load, and the back arrow is not respected, I just want to simplify to something that works and is always the same and that I can read with my old eyes.
Somewhere along the way, GUIs seemingly went from economy of use to a graphical resume for the developer (or their manager).
We didn't know as much about possible nuclear reactions back then, so I think they thought there was a possibility that there was an exothermic chain involving N or O that could be ignited by the bomb and would be self sustaining. While an asteroid impact is very powerful over large scales, it doesn't create nuclear reactions, so Trinity was indeed a first at that scale.
But, and I'm not sure how much of this they knew back then, we do get bombarded by high-energy cosmic rays, so chances are one of these hypothetical N or O reactions should've already randomly occurred at least in isolated events over the last few billion years if possible.
So Nate & Co sell out to a big corporation then are upset that it does big corporation stuff? I'm more mad at Nate here because 538 was my go-to for political coverage pre-sellout, and because of his greed it went away.
538 had some decent years through its various sellout phases and imo hit its peak after Nate Silver left - he'd long since exposed himself more as a contrarian than a serious analyst and had become a real detriment to the brand.
It was definitely sad to lose 538 how we did, but G Elliott Morris has really stepped up to continue the spirit on his Strength in Numbers blog. It's the best data-driven US politics reporting out there right now imo. He also contributes to Fifty Plus One with Mary Radcliffe, and that's excellent too, reminiscent of the old 538 polling roundup stuff that went beyond just core US politics. Recently he started a podcast with David Nir of The Downballot, which is another solid resource for lower level races.
IIRC, he got to keep all of the models and etc from 538 which is kinda all that mattered about it. Like anybody is really going to lookup the 2016 election in 2030 but they're definitely want the model's output in 2030 for 2030.
Management seems to be clearly inept and if somebody wants to give you a wheelbarrow of cash for something worthless you're generally foolish to not accept.
It's petty even for a corporation. It's not exactly maximizing shareholder value. But yeah, it's the sort of stuff corporations (especially media corporations) can be expected to do.
I think Nate has become better at this sort of stuff in recent years. But still, he said that Disney "hardly ever" interfered in their editorial process, oblivious to the implication.
(What it means is that Disney/ABC were perfectly willing to interfere in their editorial process, but rarely needed to since Nate said what they wanted him to say anyway).
538 has basically always been a licensed product, first with the NYT in 2010, and then with ESPN/ABC/Disney since they left the NYT in 2013. There were only 2 years where it was an independent blog
538 got one election right a decade and a half ago with a contrarian alternative data source, and hooked a bunch of partisan gamblers since
now you can actually bet on your beliefs and dont need to debate with anyone on whether the koolaid colored wave of choice will actually happen, or whether you are a slave to an algorithm induced hall of mirrors. note, the senate has been 50-50 for over a decade, its probably the latter
so the only intervention necessary here is on yourself, focusing on things you cant control while delusionally thinking this time will be different, over and over and over again
extracting value from partisan gullibility and pathetic power struggles is unironically the move
There's no "getting elections right"; these models just estimate probability. They gave Trump a 29% change in 2016 -- that's almost 1 in 3!
A better way to assess accuracy would be to bin predictions (0-10% chance, 10-20% chance, etc.) and see if the observed frequency aligns with the predictions.
They don't really 'call' elections since they only publish probabilities. If you'd bet on the candidate 538/Silver was more bullish than the bookies, then the only election year you wouldn't have made money would have been 2024.
That's a funny response in the context of "Now he’s more interested in profiting off of gamblers" given I remember posting with him on the 2+2 forums in the early aughts poker boom days where he was a popular poster teaching a lot of people how to profit off gamblers.
Your annoyance is understandable, but it's worth remembering: he is not your slave. He does not exist to do things for you. His providing you with something good for a time does not obligate him to provide it to you forever. Because he is not your slave.
reply