
Re: "Secure by Default"
Beat me to it.
Shame there's no BSD icon to go with Tux. Still, have a free beer.
2401 publicly visible posts • joined 16 Jun 2016
Back in the day, my generation were not allowed to take calculators or books of formulae into exams. We had to work it all out for ourselves.
My kids were required to take in scientific calculators and demonstrate that they could use them.
I understand that these days, web browsers and search engines are no longer the spawn of Satan. Inept use of Mathematica will not get you very far.
How long before the average student will be required to show competence in the use of AI?
So, firstly this is not 2D. The tracks pass over each other, so the equivalent macroscopic topology is a double-sided circuit board of the kind which was ubiquitous in the 1980s. The first CPUs used only single-sided boards, genuinely 2D and not mere marketing-speak 2D, with the first microprocessors being similarly restricted on-chip.
Group III-V (silicon-free) semiconductors were used to create high-speed devices for specialist applications.
So here we are, back to 1970/80s technology. I am not entirely clear how a selenium compound is more commercially viable than, say, gallium-aluminium-arsenide. But then, I seem to have lost my coloured pencils.
P.S. But if it can run Moon Lander, I'm in!
I would agree, except that Word has been rewritten from the ground up more than once. To save on development costs and time, MS took the opportunity to duplicate all the regular gobshite rather than improve the UX. They even had to ask the Open Source community how some of the gobshite worked, because they had forgotten. Today's Word 365 cloudy web edition is still only part way through that rise from the old coffin, and utterly unusable if you want more than coloured pencils from it. A pro tool it absolutely is not, and never will be.
Even LibreOffice is guilty: just you try to customise an Outline List in either tool. Different quirks, different borks, but both bork it nonetheless.
But it all hardly matters; as you say, the Coloured Pencil Office are unable to understand anything about it and simply apply on-the-spot formatting any way that occurs to them.
As a Tech Author constantly having to fix what our codies fondly regard as Technical Publications, I can see AI finding a home as a grammar checker and Word gobshite correcter.
Many codies are chronically challenged in this area. They have never heard of Eats, Shoots and Leaves (which they ought to, it is unprofessional for an umbrella contractor to pretend their meatbags can write when they can't), nor do they have the slightest motivation to engage with said gobshite (with which I deeply sympathise, I reckon I deserve danger money and compensation for mental torture in the workplace).
It tell them that I regard the human brain as a wet Turing machine, whose coding language is English. If you get the syntax, punctuation, code order, etc wrong, then you bork the code and the poor user suffers from GIGO. A few devs begin the long climb to grok, most do not wish to hear, which is why they buried themselves in computers in the first place and only fire up Word when threatened with instant coffee.
An AI code cleaner between them and me would be a Gift of the Gods.
Icon for the day job.
EDI, aka woke, began as a Good Thing. Zealots overdid it. The zealots on the right rebelled and turned it into an insult. Elsewhere I see the backlash against the backlash rising. Such is social politics.
Xfree86, X.org, Xlibre and the denizens of the Wayland ecology (which is no less diverse) all have their value for different requirements, and we enjoy the choice. But, being dicks, we dicker and squabble. Occasionally some fsck-ing bustard gets to be first against the wall when the revolution comes. Damn! I've run out of popcorn. Anybody want a fight in the foyer?
So those are the real memory/processor hogs, well well well. I suppose they're a great place to hide the AI and spyware so the Black Hats can't find it.
But I do think that calling the rest of the treacle-crap "productivity" stuff is a bloody cheeky spin. As Fat Freddy's Cat once said, "Well, my uncle is in medical school", with follow-up thought bubble "I'm not going to tell them what they're doing to my uncle in medical school".
I don't think they want to bugger up Markdown. They chose it precisely because it saved them the trouble.
Makes it easier for them to design their own Microsoft Markdown (TM), which will allow proprietary binary extensions to be included, data island / email image style, and then get it ratified as an "open" standard. The default for Notepad 2028, natch. I/O filters available at extra cost.
1. Hate for decades, try to kill off for decades, fail for decades.
2. Change of tactic:
a) Embrace: yay, we're keeping it folks, we hear you!
b) Extend: yay, we're making it even better - with the formatting that you all miss from Worpad, plus the latest buzzword blockchain I mean AI!.
c) Extinguish: yay, to give you an even better experience we're adding virtual colored pencils to the plain-text view, so you can still see that lovely markdown and highlight it while sharing your page on Teams. What? And we've made both OneNote and SharePoint 100% compatible* with the new version! What's that? Did we omit to mention that /editing/ in the plain-text view has disappeared? Look, if you're that hardcore, we've given you Edit on the command line, so everybody's got what we they wanted. Yay!
* Except where they aren't and never will be
In one of Asimov's Multivac stories, they hand over governance of civilization to it, heaping all the cares of the world on its shoulders. Presently, it starts to behave oddly.
The hero goes along to find out why. Evidently there is no shutdown script, for it has only one thing to say to him, "I want to die".
Certainly not through some programming bias, the origin of the phenomenon has to be buried in the data.
Maybe it's because if you are polite, the AI responds with data associated with polite conversation, i.e. politically correct but vapid burblings and engaging speculations. But smart people who try to get across anything as complicated as the time of day find themselves thwarted by those viral memes and get angry. So, if you get angry with an AI you get past the pap and are instead associated with what sensible people actually have to say.
I find it sad, as it is only going to train us up in return, for greater aggression in our personal lives, and that is not good.
Weird downvotes there. Time for a lesson in neurosicence.
Current theories of consciousness suggest a predictive model: our brains run up a model of reality based on our sensory inputs. This model tries to predict what to think - and do - next. How that affects our senses is then fed back into the model for the next round of predictions. This stream of predictions is just our stream of consciousness. This is not a bad description of the predictive algorithms - along with their training - which we brand as generative AI.
So yeah, probably more than a worm, probably short of a mouse.
And there's more.
Commentards (I plead guilty too) have often criticised LLM+tokens+"reasoning" as just predictive algorithms: that so-called "reasoning" does not (yet) extend to the human capacities for analogy or semantic/symbolic manipulation. All they do is guess which token should come next in their output stream, throw it in there and start on guessing the next token. Sometimes they guess wrong and we call that hallucination.
However, current theories on how the human mind work are also based on a predictive model. We have long known that conscious experience is based on our mental model and not directly on our sensory inputs (even the Buddha realised that all experience is illusion). The hypothesis now is that this model is just a prediction based on analysing our sensory inputs and memories; we constantly compare it with current inputs and update our predictions accordingly. The resulting stream of predictions is just our stream of consciousness. Hallucinations are what happens when the predictions are borked by drugs or whatever.
Coincidence or getting close?
Not for want of trying.
I grew up with the Model M and its ilk, and always found them a bit clunky. Too much travel, not a crisp enough feel. Always looking for better. My current has modest travel, adequately light touch, click feel better but could be better still. Gentle clicky rattle but no more. Yet it's a super-cheapo £10 membrane+keytops toy from some retail box-shifter, though one with the grace to set them all up to try-before-buy. These toys often don't last more than a few years, so I keep a couple of spares.
The problem is twofold. Punters seldom buy for UX, they buy on price, fashion and perceived quality. The market dominators deliver on that, and don't need to refine the UX. Secondly. an independent Designer with any originality or finesse will immediately change the obsolescent layout for something funkier. Of the 1,000 punters who come across the offering, there will be 2,0000 views on why it's no better. To paraphrase our Vulture: once you're used to swimming in PC Layout shit, why bother to change the water? For example, who ever sits at their desk all day, manually entering streams of numbers through a numeric keypad duplicating some of the main keys, and which is upside down compared to a phone? Lappies dumped it years ago, but desktops? Humbug! But you try to sell that through market channels.
The best mobile keyboard from way back when was the Psion Series 5. Its designer, Martin Riddiford, got almost everything right. Keys were a bit firm, and its small size meant that the biggest-fingered folk couldn't do more than poke at it like all the others, but it was undoubtedly the jewel in the crown of the PDA revolution. When Planet resurrected the form factor for the Gemini, they hired Martin to update his act too. Lighter touch, cleaner feel, truly the new gold standard. Later they went gimmicky, were pinched for cash, and thought they didn't need him. Back to the shitpond.
So yeah, there is/was a master keysmith. But nobody cares.
After 35 years of use, Google just whined at me that I now had to declare I was over 18 or it would turn shit off. That meant logging in. So, natch, I left it to turn shit off.
And it did! No more targeted advertising! No more AI summaries! Bugger all enshittification left! Almost a half-decent search engine again!
Been using Brave and Mojeek increasingly, when the Bing/Google scrapers regurgitate shit.
If this UDM14 thing catches on, you can bet the Chocolate-Flavoured-Shit Factory will find a way to re-enshittify it. So I won't be removing the above pair from my toolbar any time soon.
Mind you, the AI summary can be useful for some searches, so what I'd really like is a widget to click, which brings it up. Deffo not displayed by default.
Boom are having to build their own engines. You need supercruise - the ability to fly supersonic without afterburners - and other green/economy measures to make the engines viable. No engine maker was prepared to take a flyer (sic) on a brand new product line.
So here are Boom developing both a revolutionary airframe and revolutionary engine from nothing, flying in a higher-drag and more highly constrained regime than ever, at only around 50% faster than today but the exact same time in the stack and on the ground at either end, and expecting a small (by today's standards) 80-seat feeder-liner class type to pay its way on the long haul.
I wish them luck but, sorry, I ain't investing.
So goes the old saying.
I too have begun advocating an ecology of AI approaches, in order to achieve a Darwinian balance and prevent the risks from any one type dominating. As ever with my bright ideas, someone else got there before me.
But I wouldn't spend too much effort on malignant AI-on-AI. Especially considering the concurrent headline here that everybody is deploying AI but nobody is securing it, the superpower spooks are surely way ahead of us, with organised crime hot on their heels.
Not are, used to. Commercial payloads increasingly hitched a ride on NASA-driven Atlas and Shuttle launches, paying their way alongside the science research and other NASA mission payloads. We are still living through the fallout (sic) from the ending of that era.
Stratolaunch still want to do horizontal-to-space. But it's not as easy as it looks and they are having to build up to it in less ambitious steps, serving whatever markets they can catch along the way. This is precisely what makes them an outsider in the space race. (In passing, the mega-bird has a converted 747 flight system in each fuselage, so I suppose expanding the range of sizes on offer, without having to maintain a new flight system, must have seemed like a good idea at the time.)
Then you need to brush up on your Allied aircraft designs. The crop of NACA aerofoils which arrived in the early 1930s were widely adopted by British and other planemakers.
Just one example, here is a paper discussing the historical adoption of NACA sections on the obscure little Supermarine Spitfire flop that you probably never heard of:
https://www.aerosociety.com/media/4953/the-aerodynamics-of-the-spitfire.pdf
You must be new here.
You need to parse those claims like a Vulture. The ability to read and write the format is apt to mean read and write a subset of the whole thing, and full of layout bugs at that - and preferably to crash when an unfamiliar feature is encountered. They seldom claim to strive for 100% of the specification with 100% accuracy and stability.