mh-
mh- on Hacker News
Doesn't the AI (the computer players, I mean) "cheat" at that high of difficulty level?
I tried really hard to like Factorio, and I bounce off of it each time after playing a save for 4-8 hours for exactly this reason. I know a lot of programmers who love it, so I almost wish I could get into it.
Now I'm tempted to try again, because it feels like the kind of game I should really enjoy.
To be fair to them, in 1990 it was common enough to hear someone call any/every electronic game a "Nintendo".
There was a real risk of their trademark getting Kleenex'd, at least in the US.
Unless you've been very careful to only surround yourself with people who share your exact views, you probably do know some. They just don't speak up, because: what's the point debating it with you?
Netflix was, in fact, already using recommender models in that era.
Ratings are indeed being gamed now, yes. Not sure how long this has been going on, I've started to see it be obvious in the last 6 months.
It's very obvious when looking at the reviews for certain new indie games. All LLM written comments, most short, most low playtime.
Then actual customers come along and buy the game and over the next few weeks the game craters from Mostly Positive to Mostly Negative.
That's not what that is, nor does Steam have an affiliate program that would enable the parent commenter to make money from referral links.
The following sentence:
"This community has absorbed new processes and tools many times over its 35-year history, and will almost certainly find a way to do that this time as well. But that absorption has not always been without turbulence along the way."
You used the right term. I don't know what the parent commenter is on about.
Open source has historically meant you could download the source and build your own binary. The appropriate analogue here, IMHO, to the build->binary process is training->weights.
No idea. I used it a lot too, not sure if pasting here worked from the beginning of them introducing this Google Lens thing or not.
Click the little camera icon in the search box and it pops a modal that says
Search any image with Google Lens.
The textbox below it says Paste image link, but you can actually paste an image from your clipboard here, too.
It means the level of effort that a ChatGPT Pro plan summons when thinking. For 3 hours.
If folks are going to analyze this claim with a critical eye, I'd be zeroing in on "average" rather than acting like this measurement is somehow unclear.
That's an incredible get. As someone who grew up with SimCity (DOS), I thought I'd been hurt too many times (CS:2, SC2013, [..]) to get my hopes up again, but here we are. Wishlisted.
A surplus of outrage and a willingness to share your unfiltered reactions to things, I guess.
Why would it not matter? Everything in safety culture is making the risks acceptable. Why does ETOPS exist when we could just disallow planes with twin-engine aircraft entirely?
Wow, very cool. Thanks for sharing.
I haven't used Linear in any serious capacity, but I feel like there's plenty of market in "Jira that doesn't feel like it hates its users" and "Jira but we care about performance".
Not saying that they're shooting for either of those segments, but someone should. Plenty of enterprise orgs that need (or are convinced they need) Jira's featureset.
If the benchmark is "what's the best model to use in your Cursor subscription", why would they do that? OpenAI knew what the effects of their decision were. Hard for me to have sympathy for either party here, honestly, and I say that as someone who is a customer of both.
Are the prices on Bedrock substantially different to the rate cards of the direct APIs? Just trying to understand whether this is Opus-through-Bedrock is too expensive, or Opus is too expensive.
I think the people who aren't greeted by this dichotomy when turning their PC aren't catching the reference.
At the risk of explaining a joke: Windows tries to get you to set up things like OneDrive on startup with a "do it now" or "remind me later", there's no "never ask me about this again".
To my understanding, with batched inference and other "optimizations" you wouldn't get the exact same token predictions even with temp=0.0.
Every official statement I've seen around this is careful to say that they "don't intentionally reduce model quality", which leaves
plenty of room for "we adjusted some knobs and
our evals show performance is materially the same".
However, I also agree that I haven't seen any robust data from someone tracking it daily/weekly. The handful of sites purporting to do this aren't even running it enough times to hit stat sig.
edit: someone linked one elsewhere in this thread called AI Stupid Level - they "run 7 trials instead of just 1". I don't blame them. Doing this in a statistically sound manner would cost a small fortune.
I don't know how to respond to that, because it has quite a few negative effects on them..
As someone who was annoyed by it at the time, I don't think it's a good example of that.
This was in an era of much more limited upstream bandwidth. The p2p nature was obscured from non-technical users in some implementations. So on and so on. I'm sure there are plenty of writeups from that era detailing why it was a bad idea.
These companies already used capable CDNs to distribute patches. It was a way for them to shift their CDN bill onto their paying customers.
It was a bit challenging long before that. I ran webhosting on FreeBSD in the early 2000s - remember finding a tool you needed and hoping it was in ports?
I'm even more confused now. I guess this was an oblique way of accusing people of astroturfing?
> I don't think I have to speculate why comments like yours get downvoted so much.I'm genuinely not sure what you're implying here. Can you expand?