Slightly related, as someone who doesn’t engage in this type of work, I’m curious about the potential risks associated with discovering, testing, and searching for security bugs. While it’s undoubtedly positive that this individual ultimately became a responsible person and disclosed the information, what if they hadn’t? Furthermore, on Discord’s side, what if they were unaware of this person and encountered someone attempting to snoop on this information, mistakenly believing them to be up to no good? Has there been cases where the risk involved wasn’t justified by the relatively low $4k reward? Or any specific companies you wouldn’t want to do this with because of a past incident with them?
If you engage in “white hat security research” on organisations who haven’t agreed to it (such as by offering roles of engagement on a site like hacker one) there is indeed a risk.
For example they might send the police to your door, who’ll tell you you’ve violated some 1980s computer security law.
I know 99.99% of cybercrime goes unpunished, but that’s because the attackers are hard to identify, and in distant foreign lands. As a white hat you’re identifiable and maybe in the same country, meaning it’s much easier to prosecute you.
> Furthermore, on Discord’s side, what if they were unaware of this person and encountered someone attempting to snoop on this information, mistakenly believing them to be up to no good?
Companies will create bug bounty programs where they set ground rules (like no social engineering), and have guides on how to identify yourself as an ethical hacker, for example:
> The Paradox Mods platform will remain the only officially supported mod hub, so deep code mods akin to CS1’s may never return.
As someone else pointed out, this is false. I have also created mods for both CS2/CS1 and I would even say it's the opposite. In my opinion, CS2 allows for even deeper code mods because they have mod tooling built right into the game unlike CS1. The host of the mods (Steam Workshop vs Paradox Mods) doesn't change anything related to mod capabilities.
> ...its long-time partner Colossal Order announced a quiet but monumental shift.
Ah yes, "quiet", like how it's been posted on every CS2 social media account, and blasted in every possible space of CS2. Haha Absolutely nothing "quiet" about it.
For majority that want to switch but can’t yet, gaming is the biggest pillar still very far behind. Many popular games (and game related apps) don’t work on Linux, sadly. I don’t know if it’s ever going to change either because of it being a chicken or egg scenario where they don’t want to spend the time supporting it cause it’s not enough users, but it’s also not enough users cause it’s not supported.
> I don’t know if it’s ever going to change either because of it being a chicken or egg scenario
We don't even need native games. Proton, when it works, is amazing. Win32 is effectively now the stable ABI that Linux always needed but never had.
The real problem is kernel level anti-cheat, which will never happen on Linux, but more importantly, gamers should be pushing back against it even on Windows. It's invasive. The latest of which you can't even enable virtualization support in Windows if you want the anti-cheat to run, which also means you lose virtualization based security, no WSL, etc. It's completely obnoxious and I hope Microsoft cracks down on it, because if they do then more games will run on Proton.
> It's completely obnoxious and I hope Microsoft cracks down on it
I hope they don't. Competitive gaming has been begging to stop cheaters for a long time. Ring 0 anti cheat has shown to be very effective against the vast majority of cheaters. Compare CSGO with something like Valorant. It's clear it's effective. Is it invasive? Sure. Is it mandatory? No (sorry you just cant play the game).
Bring back private lobbies/private servers then. Make the anti-cheat optional. Those that want to play in public lobbies have to rootkit their PC or play on console, those that don't still get to play the game without it but not in public lobbies.
If anything, gaming is the pillar that is furthest ahead, thanks to SteamOS and Proton and everything else surrounding it.
The main issue is that a lot of people I know need things like Photoshop or propriety CAD apps or video editing software where the alternatives are simply not acceptable - sure I can mention some OSS alternatives but it's not really my field; this is their job and they can't really take the velocity hit, or waste time finding out mid-project that it can't do what they need it to do.
Depends a lot on what kinds of games you play, I think—I built a PC in 2020 and originally set it up to dual boot Linux and Windows, but over time I used the Windows partition less and less and wound up deleting it last year.
I realized recently that at some point I stopped even checking ProtonDB before buying games on Steam, I guess because its been so long since I've run into one that didn't work. I play a pretty wide variety of games, but not so much the type of competitive multiplayer FPS that seems to have the worst Linux compatibility due to anti-cheat.
The biggest problem is probably work-related apps not working. Adobe products, MS Office, and certain niches like the music industry just aren't supported on Linux.
Many ultra-popular games don't work due to anticheat, but some do. Dota 2, Counter-strike, Marvel Rivals, Overwatch 2, among others work perfectly fine. We've also reached a point where virtually every offline game will work too.
Agreed. All it takes is a simple reply of “you’re wrong.” to Claude/ChatGPT/etc. and it will start to crumble on itself and get into a loop that hallucinates over and over. It won’t fight back, even if it happened to be right to begin with. It has no backbone to be confident it is right.
> All it takes is a simple reply of “you’re wrong.” to Claude/ChatGPT/etc. and it will start to crumble on itself and get into a loop that hallucinates over and over.
Yeah, it's seems to be a terrible approach to try to "correct" the context by adding clarifications or telling it what's wrong.
Instead, start from 0 with the same initial prompt you used, but improve it so the LLM gets it right in the first response. If it still gets it wrong, begin from 0 again. The context seems to be "poisoned" really quickly, if you're looking for accuracy in the responses. So better to begin from the beginning as soon as it veers off course.
> The grand-parent comment was pointing out that this limitation exists
Sure, I agree with that, but I was replying to the comment my reply was made as a reply to, which seems to not use this workflow yet, which is why they're seeing "a loop that hallucinates over and over".
That's what I like about Deepseek. The reasoning output is so verbose that I often catch problems with my prompt before the final output is even generated. Then I do exactly what you suggest.
Yeah it may be that previous training data, the model was given a strong negative signal when the human trainer told it it was wrong. In more subjective domains this might lead to sycophancy. If the human is always right and the data is always right, but the data can be interpreted multiple ways, like say human psychology, the model just adjusts to the opinion of the human.
If the question is about harder facts which the human disagrees with, this may put it into an essentially self-contradictory state, where the locus of possibilitie gets squished from each direction, and so the model is forced to respond with crazy outliers which agree with both the human and the data. The probability of an invented reference being true may be very low, but from the model's perspective, it may still be one of the highest probability outputs among a set of bad choices.
What it sounds like they may have done is just have the humans tell it it's wrong when it isn't, and then award it credit for sticking to its guns.
I put in the ChatGPT system prompt to be not sycophantic, be honest, and tell me if I am wrong. When I try to correct it, it hallucinates more complicated epicycles to explain how it was right the first time.
> All it takes is a simple reply of “you’re wrong.” to Claude/ChatGPT/etc. and it will start to crumble on itself
Fucking Gemini Pro on the other hand digs in, and starts deciding it's in a testing scenario and get adversarial, starts claiming it's using tools the user doesn't know about, etc etc
I think you are confused? Tailwind is already free and open source? These are just components they sale that are pre-made to save you time. It doesn’t take away much at all from the full experience?
FWIW, no source on the cause yet. Only thing we know is from their official statement here:
> It is with more sadness than mere words can convey that we have to report that our beloved Ozzy Osbourne has passed away this morning. He was with his family and surrounded by love. We ask everyone to respect our family privacy at this time. Sharon, Jack, Kelly, Aimee and Louis
How are people using this without getting rate limited non stop? I pay for Claude Pro and I sometimes can’t go more than 5 prompts in an hour without it saying I need to wait 4 hours for a cooldown. I feel like I’m using it wrong or something, it’s such a frustrating experience. How do you give it any real code context without using all your tokens so quickly?
Try giving it a repomap, eg by including it in CLAUDE.md. It should pull in less files (context) that way. Exactly telling it which files you suspect need editing also helps. If you let it run scripts, make sure to tell it to grep out only the relevant output, or pipe to /dev/null.
I have the same issue and in recent days I seem to have gotten an extra helping of overload errors which hit extra hard when I realize how much this thing costs.
Edit: I see a sibling comment mention the Max plan. I wanna be clear that I am not talking about rate limits here but actual models being inaccessible - so not a rate limit issue. I hope Anthropic figures this out fast, because it is souring me on Claude Code a bit.
I haven't used Claude Code a lot, but I was using about $2-$5/hour, but it varied a lot. If I used it 6 hours/day and worked a normal 21 workday month (126 hours), then I would rack up $250-$630/month in API costs. I think I could be a more efficient with practice (maybe $1-$3/hour?). If you think you are seriously going to use it, then the $100/month or $200/month subscriptions could definitely be worth it as long as you aren't getting rate limited.
If you aren't sure whether to pull the trigger on a subscription, I would put $5-$10 into an API console account and use CC with an API key.
I wish there was a Max trial (while on Pro) to test if this was the case or not. Even if it was just a 24 hour trial. Max is an expensive trigger to pull, and hope it just solves this.
FWIW I went Claude Max after Pro, and the trick is to turn off Opus. If you do that you can pretty much use Sonnet all working day in a normal session. I don't personally find Opus that useful, and it burns through quota at 5x the speed of Sonnet.
> The bottom line: Apple has 24 to 36 months before it has its own AOL moment, according to Brian Mulberry, client portfolio manager at Zacks Investment.
That quote alone is hilarious. What a low effort article.
Do any of these people know that Apple is a hardware company and all of their services are just a side business? Apple is poised to make a fortune selling devices powered by their low power consumer chipsets that can run neural networks. There's really no competitor in that space.
Apple has plenty of time and money to watch other people flailing around trying to make AI first devices until they even really start to take it seriously. Even if someone else figures out to make a killer mobile AI app, it will absolutely support iPhones for the foreseeable future. All apple needs to do is make sure their chip pipeline supports it.
Google is the one that is facing an existential threat. Most people see their search engine as basically just a shitty ad-infested chatbot that produces worse results than chatgpt.
A hardware business cannot demand a P/E ratio in the 30s range, especially given that supply chain disruptions are going to eat heavily into their margins.
That's not an accurate quote. You skipped the rest of it:
> "With the cash they have on hand and the loyalty they have…there would have to be something disruptive in the marketplace that would draw away customers. It's not there yet," he says.
I take that to mean they're in a good position now, but they might start losing customers in the next 2-3 years, should stronger competitors show up. I don't disagree. I don't see Apple doing anything special that will protect them as different kinds of hardware come to market. Steve Jobs was responsible for the iPhone and iPad. Apple can only ride on his work for so long.
I really hope investors buy into this as it’ll be a great new entry point into the stock. I mean they seem to be buying into whatever without thinking so why not. AOL moment indeed.
Slightly related: For all the crap the iOS store gets for many (good) reasons, this is one reason I actually LOVE to buy subscriptions through iOS/Apple when that option is available for a platform. They have the most simple cancellation process to manage all your subscriptions in one place. Sometimes it costs a $1 or more to buy through iOS but it's worth it to easily cancel without any hoops.
My uncle got a surprise $100 credit card charge. He had clicked on a scam pop-up on his Apple iPhone. Somehow that led to installing an app and paying for an annual subscription. One support message to Apple got a refund.
Absolutely! I’m at a stage where I prefer to buy a subscription via Apple. It’s so simple to cancel. Not to mention how quick Apple is in general refunding any of the wrong/unsatisfactory purchases.
