Agents md file doesn’t actually do anything except using more tokens https://arxiv.org/pdf/2602.11988

LEDs themselves are quiet but some SMPSs generate RFI. You can find SMPS that is RF quiet, or since LED strips take so little power, you can make a tiny linear power supply for them, linear power supplies don’t have RF problems

Moxon is very slightly shortened 2el yagi. That other thing is effectively weird phased array of two almost halfwave dipoles, and it’s both worse and bigger, and also harder to make because it requires capacitors for gamma match

(Current rabbit hole: Should a HB9CV’s boom be conductive and electrically bonded to its elements?)

Yes it has to be, that’s how you feed its two elements. Boom is at zero electrical potential and forms half of transmission line that goes to gamma matches of both parts of antenna, and you can attach it to perpendicular conductive mast, or extend it for mounting to mast (has to be perpendicular if mast is metal)

Personally, i’d go with 4el yagi if we’re talking portable operation, because boom length is such that you can make it out of plastic tube and hide elements in it when moving around. Or moxon, if size of 4el yagi is prohibitive. Moxon gets you 2db more gain than HB9CV in similar footprint

openhamprep explanation is wrong

When you consider a segment of wire, it has some inductance and capacitance in that small segment, and the wider it is, the lower inductance and higher capacitance are. In other words, both X_L and X_C go down as diameter increases. Electric field lines within antenna are mostly perpendicular to wire everywhere except on ends, where they spread in all directions, which means that there’s some residual capacitance there that needs to be charged each cycle. This is called end effect and the wider wire is, the more of it you get. In fact, the same aspect ratio (length/diameter) gets you the same end effect ratio each time, and it’s usually in range of 0.98-0.95. If there’s insulation, then insulation has lower velocity of propagation as it has higher permittivity than air, and it is in place with highest electric field intensity, and this can also shorten necessary wire length a couple %. The more insulation there is, the more shortened antenna becomes. have a calculator: https://www.translatorscafe.com/unit-converter/en-US/calculator/dipole-antenna/

But wait, there’s more. We can think of antenna as a lossy RLC resonator (sort of), where R is Rrad which for halfwave dipole would be 72 ohms, + loss resistance which is smaller, and neither are changing fast with frequency. What is changing faster is X_L and X_C, and the wider wire is, the smaller both of them are (X_L = -X_C at resonance), and this means that you can go further off frequency and still have acceptable SWR, i.e. antenna made with wider wire has more bandwidth, or if we look at it as resonator, lower Q. For HF this means using masts or wire cages instead of single wire, but for VHF this gets more practical

Resistive part of antenna impedance near resonance depends on physical antenna length, and if antenna is shortened, then how it was done, but for the range we’re here it won’t change much. But also it will depend on height above ground, type of soil, antenna geometry and many other things. 72 ohm is in freespace only

you can consider antenna wire as a transmission line, but it will need to include ground as the other side of transmission line, and then impedance is dependent on wire diameter to height over ground ratio. it’s not very useful except when considering very large and currently uncommon types of antennas (beverage antenna and rhombic antenna, both of which are traveling wave antennas). other than that, transmission line interpretation only works when antenna is very close to ground in wavelength terms, and usually it’s way below what you need for radiation to work properly. when it’s way up there, or in free space, there’s another impedance that is specific to wire in free space only and not dependent on length, but it’s not very useful for explanation of anything in antennas

careful, they might like it

humans start breathing only when 3 years old, don’t you know?

New-ish EU national IDs have NFC capability and certificate in them so you can probably sign something with that if you figure out how

lol nope they can’t do that because “guardrails” aren’t anywhere near reliable, and they won’t because it would cut into their profits userbase numbers, based on which they raise vc money. delusional chatbot user is just a recurrent subscriber

so they’ll allow uploads of papers newer than 2020 after the case is closed (either way)?

llms allowed them to glide all the way to the point of failure without learning anything

maybe check from openai didn’t clear

if it’s permanently installed high power device, it only makes sense to wire it directly, like it’s done with ovens, and EV charger draws even more power. why there’s even plug on the way there? it’s not gonna be moved anyway. and while we’re at, why you people don’t have three phase circuits for big loads like this? standard euro five wire three phase 32A circuit (or plug) gets you 22kW, that plug tops off at 9kW

80000 hours are the same cultists from lesswrong/EA that believe singularity any time now and they’re also the core of people trying to build their imagined machine god in openai and anthropic

it’s all very much expected. verbose nonsense is their speciality and they did that way before time when chatbots were a thing

bro tried to recruit jihadists in roblox (failed) and now screenshots from this all are a matter of public record 💀

For slightly earlier instance of it, there’s also real time bidding

taking a couple steps back and looking at bigger picture, something that you might have never done in your entire life guessing by tone of your post, people want to automate things that they don’t want to do. nobody wants to make elaborate spam that will evade detection, but if you can automate it somebody will use it this way. this is why spam, ads, certain kinds of propaganda and deepfakes are one of big actual use cases of genai that likely won’t go away (isn’t future bright?)

this is tied to another point. if a thing requires some level of skill to make, then naturally there are some restraints. in pre-slopnami times, making a deepfake useful in black propaganda would require co-conspirator that has both ability to do that and correct political slant, and will shut up about it, and will have good enough opsec to not leak it unintentionally. maybe more than one. now, making sorta-convincing deepfakes requires involving less people. this also includes things like nonconsensual porn, for which there are less barriers now due to genai

then, again people automate things they don’t want to do. there are people that do like coding. then also there are Idea Men butchering codebases trying to vibecode, while they don’t want to and have no inclination for or understanding of coding and what it takes, and what should result look like. it might be not a coincidence that llms mostly charmed managerial class, which resulted in them pushing chatbots to automate away things they don’t like or understand and likely have to pay people money for, all while chatbot will never say such sacrilegious things like “no” or “your idea is physically impossible” or “there is no reason for any of this”. people who don’t like coding, vibecode. people who don’t like painting, generate images. people who don’t like understanding things, cram text through chatbots to summarize them. maybe you don’t see a problem with this, but it’s entirely a you problem

this leads to three further points. chatbots allow for low low price of selling your thoughts to saltman &co offloading all your “thinking” to them. this makes cheating in some cases exceedingly easy, something that schools have to adjust to, while destroying any ability to learn for students that use them this way. another thing is that in production chatbots are virtual dumbasses that never learn, and seniors are forced to babysit them and fix their mistakes. intern at least learns something and won’t repeat that mistake again, chatbot will fall in the same trap right when you run out of context window. this hits all major causes of burnout at once, and maybe senior will leave. then what? there’s no junior to promote in their place, because junior was replaced by a chatbot.

this all comes before noticing little things like multibillion dollar stock bubble tied to openai, or their mid-sized-euro-country sized power demands, or whatever monstrosities palantir is cooking, and a couple of others that i’m surely forgetting right now

and also

Is the backlash due to media narratives about AI replacing software engineers?

it’s you getting swept in outsized ad campaign for most bloated startup in history, not “backlash in media”. what you see as “backlash” is everyone else that’s not parroting openai marketing brochure

While I don’t defend them,

are you suure

e: and also, lots of these chatbots are used as accountability sinks. sorry nothing good will ever happen to you because Computer Says No (pay no attention to the oligarch behind the curtain)

e2: also this is partially side effect of silicon valley running out of ideas after crypto crashed and burned, then metaverse crashed and burned, and also after all this all of these people (the same people who ran crypto before, including altman himself) and money went to pump next bubble, because they can’t imagine anything else that will bring them that promised infinite growth, and they having money is result of ZIRP that might be coming to end and there will be fear and loathing because vcs somehow unlearned how to make money

congratulations on offloading your critical thinking skills to a chatbot that you most likely don’t own. what are you gonna do when the bubble is over, or when dc with it burns down

chatbot DCs burn enough electricity to power middle sized euro country, all for seven fingered hands and glue-and-rock pizza