1,059 karma · joined August 31, 2014
Love the image. "Hey you! No visual aids go into this town! One more step, and my strawmen will React!" But the accountant with the abacus continued towards the east gate. He squeezed between two strawmen, who didn't React as much as the gatekeeper was implying.
I had to visit the site of whatever service provider they use. I had to login with 2FA and accept their terms. The terms allow third parties to send me targeted advertisement based on my medical history. (They are very explicit about that. I guess that's progress... This is in Europe.)
So the stuff I get "for free" when I sell my medical history is the right to view my own medical bills. If I don't agree, I just get charged some random amount via my health insurance. I could bother them to share their copy, but by default they don't.
What you're looking for is the "anti-features" distinctions, like the F-Droid store does it. E.g. "This app depends on other non-free apps." or "Promotes or depends entirely on a non-free network service."
I thought it was a good choice. They were assuming deep familiarity with modern CPU architectures through the whole text, so a one-sentence definition of all the terms would either have been too much, or not enough, depending on the reader.
It's probably the same issues as with neural networks: they have to work on a whole batch of inputs to be efficient. They load only a few weights at a time into registers, then use those on the whole batch before loading the next weights.
So my guess would be: If you only run a single grid, and the grid is large, you may end up memory-bound and you can afford running inefficient instructions.
(Btw. sounds like bitgrid would map 1:1 into the LUTs of an FPGA. Might be a fun project to generate VHDL for it and then let it actually run in parallel. The less straight-forward part being the logic to get the data out of the FPGA again.)
To remedy to that, I highly recommend Joseph Henrich's "The Secret of Our Success". What we call "human behaviour" in such discussions is often culturally learned (or at least shaped).
His point: our psychology is "programmable", and prestige plays a key role in that programming. Western culture (market norms, trust towards strangers, monogamy as the norm, etc.) is forming our psychology in a way that is very far from (historically) typical humans.
(That said, I'm no expert in this area and cannot put Henrich's work into context.)
E.g. when you do summation using a for loop, and the compiler is trying to preserve the rounding-correctness your float operations. Or because you created a dependency graph with a for loop where each iteration requires the previous iteration's result.
If you, say, get 80% of your salary after you lose your job, but lose that benefit when you start working again, then you are effectively taxed 80% of your new income. So financially, you don't have much of an incentive to look for work. An alternative would obviously be UBI, where you may earn less (because of general higher taxes to pay everyone an UBI), but what you earn in addition you can mostly keep.
Like asking someone who prepared a speech to smoothly skip all words with "ei" in them. The LM would basically need to have memorized the set of letters that each of its token stands for, from the training data. I'm surprised it works at all.
Depends on your definition of "intelligence". The big missing part is the ability to explore, try new things, to act (enactivism). Basically to become part of the environment, instead of being a sealed box with frozen weights.
By predicting characters, the system had to master, digest, maybe even understand, all the cultural human knowledge it got in text form. Now let's aim for the process that generated this knowledge in the first place.
After that, it's dolce vita and stargazing until they die.
More seriously, I don't think speed-of-light has much biological relevance. It has consequences for microchips, though.
Not sure about calling this "real time". If you actually wanted to do something about it (like saving the poor, mysterious object), then any help you send will face a completely different situation from what we see now.
FreeCAD is like entering values into a form, while Blender is like playing around with the model interactively. You can drag anything around, tear it apart, fix it again. It's possible to keep things editable, almost parametric, but this requires more planning and knowledge.
For getting started you definitely want to follow tutorials. One for basic mesh modelling, and one with focus on CAD. (How on earth are you supposed to figure out that you want a "subdivision surface" modifier. It's the first thing I add in every project.) But I'm thinking now FreeCAD may be just as hard to learn as Blender (unless you already know CAD).
(But now that I know the basics of CAD workflows, CAD Sketcher looks interesting. Before that, I wouldn't have seen the point.)
So your post about keyboard layouts or databases will more likely be discovered/boosted (even if you have no followers on other instances), because the local timeline and hashtags are more quiet than when you're on a general instance (where your posts drown in the noise). Or compared to, say, sauropods.win, were you more likely reach a geologist than a database admin.
Generally I like to be on two or three instances, so my art doodles go to mastodon.art rather than a tech instance (flooded with AI art). I like to check the local timeline or discovery page of other topic-instances via web occasionally, maybe follow someone there to peek into another bubble. The big instances like mastodon.social are just too much, flooded with random stuff I don't care about.
About monitors and larger stuff, you may argue. Employers can find and buy good ones in bulk, instead of making everyone individually spend hours being annoyed, searching and comparing and buying better equipment. Either way, for monitors the financial hit will be way less than 1% of a typical salary (okay, may depend on location, but a monitor lasts 5+ years).
I agree it has to stop somewhere. For me that would be stuff like Unity3D subscriptions, or tables, or FPGA tools and oscilloscopes. But monitors... shrug