Qwen 3.5 is light years ahead of llama 3 and deepseek, but no comparison to Claude Opus 4.6. Sorry. Plus: the full 35B model requires either a massive GPU (in the multi thousand $ range), or at least a lot of RAM (which is currently a bit pricey). Either way: I have the strong impression, that OpenClaw will lose quite a few users over this.
But you've got to do both. Doubting oneself is "critical thinking". Doubting other sources of authority is "independent thinking".
The thing is, nobody has enough expertise to be an independent thinker in every area. So you essentially MUST delegate your ideas in some areas (variable between people) to external authorities. At which point what you "believe" depends on which authorities you choose.
A related question is "how firm is that belief?". This also tends to vary wildly with little apparent (to me) reason behind it. This is one feature that *can* be related to IQ, but isn't always.
Sorry to rain on your parade, but qwen is no match for Anthropic's premium models. I've used both for coding relatively easy stuff, and qwen 3.5 puts lots of bugs even into three page shell scripts, while Claude's code can often be taken as is.
Why does this matter a lot? The biggest threat against lobsters is "prompt injection", and only top of the line LLMs are moderately resistant to it. Running an OpenClaw install based on an entry level LLM can be very risky once you give it access to passwords or personal data.
Anthropic evidently knows this, that's why they see little risk when they massively jack up their prices. Those dropping out now were never the type of people who would ever throw significant money at the effort.
It's not just widespread, it's universal. What varies from person to person is the domain that they apply thinking to, and how they validate the authority they choose to trust.
Nobody is an "independent thinker" on every topic. Wherever one is an expert, one tends to be an "independent thinker" in that domain. Where you don't feel knowledgeable, you tend to accept an authoritative source...possibly after doing some amount of checking to see whether others think it reliable.
I don't think it's directly related to IQ. I also don't think it's restricted to chatbots. A lot of people are willing to accept the opinion of any authoritative source that they've accepted. Think religion or political party. Once they accept it, they stop questioning it's proclamations.
Note that this also applied to those who accept the proclamations of scientists or compilers. Once you accept an authoritative source, you pretty much stop questioning it. It's been multiple decades since I really argued with a compiler...unless it was a known bug from a source I trusted. I generally just assumed that I misunderstood what the language meant by that construct. (Of course, the few times I really didn't accept it, I eventually turned out to be wrong. Oh.)
Yeah well, we can't all be a person who doesn't understand how things work. Kudos to you for being simple minded!
" The Kathmandu Post"
Ah we don't to pay for journalism, says a planet of suckers
The UK has reliable rainfall data from the 1860's (165 years), and less reliable records going back further. But compared with 1,200 years, you have the right magnitude.
Oh, that's pretty neat. Microsoft is definitely the right level to address this at - they already have permission to enumerate the HW, own the hardware and software infra to tackle this, enjoy economy of scale other players are not privvy too, and can deliver a solution in a vendor agnostic way. Thanks for the heads up. It's the right thing to happen.
Of course there are. Tragedy of the commons. My point is that no single entity is likely to absorb the costs unless they're already enjoying economy of scale advantages and there are business experience/optic benefits to doing so. The poster above you pointed out that Microsoft seems to be addressing this, which makes a lot more sense to me than doing it at the 3d HW vendor level.
Sure, but many people would opt in, especially if you explained that they would benefit.
Maybe. Maybe not. Before committing to developing such a thing, you'd have to at least do some research and analysis to find out if that's true and how the likely opt in/out ratios would impact the business case. Remember, this is hosting content in a daemon on your machine
It can't be only when the game is open - this is when gamers are most sensitive to their computers doing other work, and the available of such a network would be far more limited.
Also a torrent like network would be absolute loaded with cache misses. You need to fetch a shader from somebody who has the exact same hardware/drive/game version combination as you do, and they need to have opted in. I highly suspect the majority case for many would be to cache miss and end up compiling locally.
Asking people to host and serve a non-trivial amount of content to other players is a non-starter. (The size of compiling all the shaders for CoD can range from a couple gigs to 10 gigs.) Opting in to a torrent-like network would have to be opt in - many people would just opt out (justifiably or not) minimizing the point of such a network.
You can probably assume that if you've thought of something, they've thought of it too. They simply have constraints and considerations - both technical and business oriented - you don't need or want to account for.
It's worth noting that many game studios/engines do support shared shader caches in their local studio pipelines, but the hardware config spread is much more limited, and the costs for lost productivity waiting for shaders is far greater than hosting a shader cache on premises.
"Little else matters than to write good code." -- Karl Lehenbauer
