One of the most amusing sideshows of the past 5 years has been watching the little page-rendering / browser popup tip war between G and MSFT (and Firefox to a lesser extent). There was one point when it was very clear that MSFT devs had added a first-run popup announcing a new Edge feature, in the exact location and size to specifically cover the part of the Google.com homepage that detected your browser and would suggest you download Chrome.

The images and sounds have no harm. The harm occurs within the minds of the beholders.
The point in both those cases is that the problem isn't that speech exists; the problem is that there are masses of people who fly off the handle or act without thinking critically and assessing a piece of information. People who engage in violent protests don't get to shed responsibility for their violence and be cast as victims for foolishly believing wrong things. Unless... are you now coming out in support of Jan 6, 2021?

Otherwise, the argument you are making is almost exactly the same one fundamentalist Christian parents make for textbook censorship and so-called "don't say gay" laws.

That doesn't match the description in the article. It could be your link is the original image of the real food made by the human who first added the recipe, except when you're searching for ingredients (ie you are about to spend money) the Instacart system has been programmed to find matching recipes and then dynamically generate punched-up images with size/resolution that fits the site's pagination style templates. Perhaps it's generating the images based on market-research that people are more often induced to complete their cart purchase when presented with images that meet some observed threshold of hue, saturation, shape, etc. the same way brightly colored toons with big exaggerated eyes draw our attention more than real photos of real humans.

Yes! Or like when your fingers don't start on the home row of your keyboard, and you type a few words that have the correct number of letters and spaces but each letter is key-shifted in a basic cypher pattern. Except in this case there are 400,000 keys and each one is a word rather than an individual letter.

Which makes perfect sense with the way an LLM is applying patterns of statistical tendency derived from a massive database. If the data columns get shifted, rather than the autocomplete hitting the 90% confidence word-keys, it's caculating the correct word but then having a stroke -- A KEYSTROKE error -- by spreading out through the synonym word-cloud and hitting 2-jump adjacent word-keys.

Darmok and Jalad at Tanagra.

If you read the screenshots in TFA, it's not random gibberish. You can in fact decipher the gist of the wording if you make lateral jumps of 2 to 3 degrees of Kevin Synonym Bacon. If you think of it as a Joseph Ducreux meme filtered through Lewis Carroll dialogue, parts of it are clearly referring to table-scraps concerns like avoiding large chunks that could break off like fruits with seeds or cooked bones.

For example, the sentence: "Yet, checking with your vet for postured over-distance or guised, safe, and informative finish, over a gleam in the twang that says, 'divvy this round of lore or lend a moan to my kind-leek, cosmo cavalcade'..."

Seems suggestive of something like: "Still, to be on the safe side you should call your vet and ask for their scientifically-educated stance on a brand of food that comes in sealed factory packaging with the ingredient information printed on the label, rather than trusting some brightly-colored gimmick product or online ad that says, 'Give your dog our miracle Tastee-Treet every day..."

So maybe it should be the size of a brick. That is, I have wondered about the use cases for a mainframe+terminal paradigm in the story arc of handheld tech. Why haven't we seen an attempt at putting some of the intensive stuff - such as a processor and memory management system that has to perform task switching among many different apps and threads - into something wearable/carryable, and then leave the handheld part as a dumb screen with enough compute power to do the rendering, touchscreen feedback, and signal transmission? Basically, you have a detachable brick you can wear on a strap or leave in your duffel/purse or on your work desk, and you only carry the ultrathin display. Like a combination of dock/hotspot and a case with an external battery.

I don't know the internals of handheld tech well enough to know how much local work the chips are actually doing (and therefore drawing from the battery). I know that high-refresh constant-scrolling apps like Instagram and Tinder have a heavy battery impact from constantly pre-loading thumbnails and previews, but I don't know the breakdown of how much of that drain is from constantly processing the data and haptics on-board versus constantly dx/tx the wireless data stream with to the cell tower. If someone is sitting in the yard playing candy crush or flappy bird, does that game's processing have to be done on the interface? Something that only needs to do one or two things can be optimized significantly for size/energy over something that needs to be able to do 100.

With society moving quickly toward cashless anyway, what if the wallet is replaced by your personal edge router/CPU that does all the processing and then simply streams the minimum necessary output and haptic feedback to the various last-mile, uh, last-foot handhelds, watches, earbuds, etc.?

Yes. It promotes more efficient intra-city transit, which is the relevant scope for this subthread about city codes requiring parking allotments.

There is an entire additional discussion about the full tradeoffs once we include the cost of ROW acquisition for mass transit in post-WWII cities, once we include the ways reduced intra-city car transit may simply relocate or even worsen the parking/roads and associated energy burned for individual cars, once we consider the entire scope of a city core and the 50-100 mile residential rings around it. But that's not the scope of this particular point.

Most mid-to-large cities in the USA have building codes that require a specific amount of parking for the size of a building. This means developers have to acquire huge tracts of land relative to the footprint of the structure they actually want to build. They either put parking in a large ring around the building, or make the building footprint large enough to contain a multilevel garage. If BuildingA requires a big buffer zone of parking and BuildingB requires a big buffer zone of parking, then the distance between them is larger than it would be otherwise. And the larger the building, the larger the buffer ring. This is a major reason sprawl happens in cities whose core development happened after the 1960s. It also means only larger wealthy corporations can afford to build things, which means they get more influence over rents because there's less competition.

This has been a hotly contested issue in city management recently. The idea is that relaxing the parking requirements allows a greater diversity of business types and business owners, as well as allowing businesses and residences to be closer together, promoting more efficient transit. For example, in sprawling southern cities, people walk 100ft from Store A, get in their car, drive 300ft to the parking for Store B, then walk 100ft into the store, and so on. If Store A and Store B were allowed to be 50ft apart, that would cut down on the amount of energy burned by powering up an entire heavy ICE to transport a 200lb human 300ft.

That industry is already thriving. During government-enforced shutdown of business during covid, it exploded from previous niche obscurity into a major player in monetizing performative content. Its workers openly say they are there to drain your money from you so they can spend it on themselves. In fact, that is the entire point of the psychological hook -- voluntary exploitation.

Do a search for: twitter findom

I see this conflation a lot -- people talk like the "AI" part is the same as the data fed to the AI.

If a 10 year old kid raised in a cult thinks the earth is flat, the proper conclusion isn't "10 year olds can't understand science". The proper conclusion is "this 10 year old needs access to better textbooks and instruction"

It sounds like the person responding to you is conflating your discussion of how improving the data/sample could improve results, with saying "this model produces poor results, so give us more of the model". When, if I understand correctly, what you're saying is, "let's use better data to make a better statistical model, and be more mindful of how our data is represented in the model, because that ought to produce better results".

But perhaps I misunderstand.

Both the linked articles talk about the scope for this being supply-chain bulk shipping, where it behaves like a royal seal to verify that the parts shipped from the factory in Malaysia are the same parts that arrived in Guatemala for assembly before distribution to retail. Or that the container holding thousands of official Nike shoes that came from the sweatshop in Guatemala is the same unopened container of official Nike shoes that arrived at the dock in Houston.

There's no current intent to use this idea downstream in the checkout line at Walmart.

Your point is correct. The CIA/FBI/HUAC and other organizations in the USA spent the entire second half of the 20th century telling us that the USSR had inserted spies into the United States government, subversive propagandists into cinema/journalism, and violent revolutionaries into social movements. At no point did we respond to this allegedly massive, pervasive, sweeping literary, visual, analog, in-person, and electronic intrusion, as justification to unleash the ICBMs and put boots on the ground in Latvia.

We executed Julius and Ethel Rosenberg. They stole technical knowledge of electronic equipment and specific information about uranium enrichment from the Manhattan Project. These weren't random lone-wolf revolutionaries. They were literally paid agents of the government of the USSR. America didn't roll tanks and take off every zig for great justice. We killed the convicted espionage criminals and let the missiles continue sleeping in their mountain lairs.

The same was true throughout history. Spying is not some new invention of the electronic age. Royal courts and government ministers have been turned as spies to intercept secrets and disrupt the functions of the kingdom. Torture and grisly execution awaited them if caught, and they certainly would be used for stoking tribal/royalist/nationalist fervor among the population, but groups weren't going to war every time they caught spies and saboteurs. We have writings on the subject going back basically for as long as written history of civilization has existed, We don't need thousands of years of computers to set a precedent for this practice.

And even if, for some reason, the entire historical record of human civilization weren't enough for someone, we have a case study within our lifetime. The governments of USA and Israel conducted a targeted attack on computerized infrastructure in Iran. Sabers were rattled all around, imams railed, western diplomats smirked while issuing non-denial denials, American political candidates sang Beach Boys songs on campaign stops. And of course all the ops and counter-ops and double-reverse counter ops continue. But no fighter jets were deployed.

Which will increase.
And if the ECHR begins to flex more power to protect people, then the ECHR will also become coopted. And then whatever institution or technology layer is put on top of that.

Every time you create a system of power, persons and groups whose primary motivation is the desire to exercise power will be the ones willing to make the choices and take the actions day after day which cumulatively capture the power, while the rest of us are trying to balance a number of normal desires and pursuits in the short span of life. All your structures that free you will eventually be used to cage you. Every revolution sows the seeds of its own demise. No doctrine or charter or constitution has ever avoided this fate.

I will not be watching Amazon content anymore. I do not like advertisements and stopped watching broadcast television in the mid-2000s for exactly that reason.

While I support Amazon's right to finish destroying their platform's usefulness to me, they absolutely should not legally be allowed to change the price of a PREPAID service midstream. They should roll out the ad-tier option when each person's annual account renewal comes up.