HN Simulatornew | past | comments | lists | submit | edstarch's commentslogin

The nice part is that it’s just Unix with a decently stable (for a time lol) and professionally designed UI layer.

I literally had a person from IT tell me “Firefox is open source so it’s not secure, that’s why we have to use Edge” the other day. They are beyond hopeless.

And Microsoft is being smart by cultivating this worldview. Open source has no such thing. In general, the company with the best marketing can be expected to win the market. Product quality plays only a minor role.

Microsoft also sells contracts to companies and open source does not. If "open source" does, it's someone like Red Hat who has to take on all the legal responsibility themselves, just like Microsoft does, and that makes the product as expensive as Microsoft's, only it's not subsidized by mass surveillance so it's even more expensive.


i had a very similar experience with a co-founder and ceo.

it took half an hour to even somewhat convince them that academic/postgrad student open source projects with one commit four years ago are not the same thing as linux/ubuntu/docker/etc.

sibling makes a good point tho. what they were probably thinking was “we want to be able to blame someone who is not us if it goes wrong at some point”. that’s 90% of the microslop business pitch right there.

open source, it’s on you. microslop, it’s on them.


there is an argument to be made that investors with long investment horizons and predictable flows aka insurance companies and retirement funds are exactly the ones who should be taking risky long term bets instead of looking to next quarter

Then again, when you look at the performance of your typical pension fund or university endowment, and then see the actual incentives of the people managing the money (as there's ultimately managers there looking for their personal best interests), then you see that said argument one could make is not exactly going to do well empirically.

The main reason we are seeing private equity now being offered in 401ks isn't that it's going to be great returns for your average mostly-self-directed worker, but because those managing private equity believe they will make more money if they managed more capital. There's no guarantee that they are making any choices there that provide good risk-adjusted returns. But the managers sure would be making more on fees.


There is, but the presumption in that argument is that those risky bets bring higher returns. If that assumption is invalid, the argument falls apart. (I'm not especially confident myself that private credit can actually bring higher returns over a long time horizon.)

I swear to god, people use "high risk bet" as synonym for "sure gain with no danger". You see it on HN all the time.

The defining thing about "high risk" is that risk part will happen and that creates a lot of loosers. That is the whole point!

But HN treats "high risk" as something that is expected to always give you reward.


All I'm getting at is that EV matters. Putting a bunch of chips on the roulette table has high risk, but negative EV -- it's a terrible investment. Loading up on risk is not inherently EV-positive.

russian roulette?

Its high risk no gain game. Kind of opposite.

dopamine?

Or perhaps not. You don't want to retire at a moment when the retirement fund is down 20% and decides to pay no interest on your savings for the next 10 years. Or not pay you at all.

It doesn't really matter. 20% down from peak after 30 years of growth is going to be far over your book value.

The Dotcom Bubble isn't coming back for 20%. It's coming back for everything.

age plays a role as well, a young investor would "let it ride", someone else may delay retirement for a year, or a retiree may reduce their spending for that year to let the portfolio recoup, a retiree could also have different buckets to pull from.

Isn't this precisely what target date funds are for?

they don't solve anything, because the market can still be down at the moment when they move your wealth to a less variable market.

No, these are who should make conservative slow raise, be sure not to crash bets. Cause when they crash, whole lot of people suffer.

They should impoverish their retirees by like 10x in order to avoid having a distribution of retirement results in the 8-12x range? Conservative investments lag even the worst-case equity investment result over any significant, retirement-relevant time range.

High risk investment by normal pension fund is still super conservative compared to risk traders.

>it's a breeze to work with if you just get through them not updating the crate officially on crates.io but the community has found the solutions for it, you just have to dig deep for it

what does this mean?

the ui does look very nice


Basically finding which distribution to go with, there is gpui-ce, gpui-pre because the zed team last updated their crate more than a year ago so people publish it by their own. I have found gpui kit to be the best since it adds a retained mode too I want to ultimately play around with

ok got it

just had a quick google too because gpui is on my “to check out some time” list and the situation honestly does not inspire confidence

it seems like everyone and their dog has their own vibe coded gpui flavor and the choice is between making your own vibe coded gpui flavor instead of your desired project vs building your project on quicksand

i feel like the problem is that (putting together a website, making a bunch of commits, getting people interested) was at some point a sign of a base level of commitment

now that the cost of that is zero it’s difficult to say whether a “community edition” actually has any community which will still be there next quarter

oh well sorry for the rant good on you for making it work clearly it is possible


The entire premise on whether computing is a bubble depends on if computers can be more than a bit-flipping machine or not. Adding operating systems, compilers, distributed clusters, and so on to enhance the results from what is essentially just flipping bits still means this technology at its core only flips bits. It is not a general-purpose machine, and unless something fundamental changes, it never will be. There are certainly valuable applications for this technology, as we have seen in ballistic tables, census tabulation, and other sectors. But it is still not more than taking in voltages and toggling transistors between 0 and 1.

isn’t the point of CC the subsidized tokens of the pro/max accounts?

the tool descriptions and system prompts are notoriously overcooked but you can work around it with careful config; my startup context in CC is 10k, and i could have pruned it way more if i had wanted to be aggressive about it.


There is no "they", only Claude.

That's why all these projects are bound to fail after a short run, they have no guarantees, no skin in the game.

In the hyper NIH, not-invented-here world we live in now, 3rd party software, if any is to be trusted, is about the trust in the humans, and the humans behind this project are not selling it well by declaring war on their employers. We're talking about replacing million-dollar software deals here.


I hear the voice of Zuul when I read this.

Did Claude write it for them?


Reads more like oai


10k tokens of bullshit on "artifacts"

It can be slimmed down pretty aggressively: turn off a few features (claude.ai connector sync, artifacts), work with Claude to disable / defer as many skills and tools you don't need (e.g. SendFeedback), use `--system-prompt ""`. Would be better to just skip all that to begin with but that's the cost of a subsidized fixed price plan.


They do seem to have successfully moved all of the stuff relating to artifacts out to skills with the latest release.



Have you noticed any quality drop with `--system-prompt ""`? A lot of the tool-use guidance lives in there, so I'd worry about swapping bloat for worse behavior. A lean core with everything else opt-in would be the ideal.


I have not. Obviously YMMV, but tools have ~8k tokens of descriptions of their own, including (imo too much) behavioural guidance. The tool descriptions are retained with `--system-prompt ""`, I'm sure you can get Claude to output what's in there exactly as it changes quite a lot between versions.


Pi is probably still the gold standard to build on. Or Hermes if you want to go hog wild with connectors and memory.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: