31,950 karma · joined January 19, 2013
EDIT: To be clear, I'm talking about the "Written in Morse Code" example, fully hallucinated text. The AI agents seeing a decoy message isn't as bothersome to me.
A lot of car shops with regular 100% humidity conditions will swear by dot matrix + tractor for feeding paper + printing. Plus, the carbon copy forms are guaranteed to be exact carbon copies which also leads to legal guarantees about copies of paper being provably exactly the same in the court of law.
Costco shoppers buy a lot at a time. Because Costco forces you to buy 4 tubes of Toothpaste, 24 eggs (or 60 eggs), minimum 1 gallon of milk (no half gallons or pints), and like 20 lbs of rice / 10kg for the Europeans who havent been here and like 3000 meters of plastic wrap.
For Costco, the efficiency is the shear size of the shopping carts and shear mass of the goods sold at a time.
------
You literally can't buy only one bar of soap or one toothbrush like you can from Amazon.com or other stores. There's efficiency here because of simple mass.
In contrast, I can look out and see the Amazon packages in my neighbors doors. It's all single items across the neighborhood.
The vast majority of those Amazon packages are for one thing. When the inefficient pickup truck comes back with a whole weeks worth of $200+ groceries, that further increases the efficiency of the home buyer.
It's unlikely that a daily commuter would go to Costco for just one gallon of milk or a few batteries. But I know from my Amazon deliveries that single items are delivered all the time.
-------
Anyone grabbing just some extra milk or toothpaste is likely grabbing it at an even more convenient store, like 7-11 (mostly because you can't buy one toothpaste at Costco lol).
No one is driving an hour out of their way for groceries.
And even the F150 truck example: if they are driving 30 miles to work, but 10 miles to Costco and 25 miles to home (Costco being 5 miles out of the way.), that F150 going 5 extra miles is more efficient than a Prius driving 25 miles from a Costco to their home.
Integrating routes throughout the day that matches your driving habits is a basic adulting task that everyone does, and has reasonably high efficiency.
I honestly can imagine that Costco is overall more efficient than Amazon, especially for people who do shop at Costco. If there's no Costco closeby, its more likely that the individual humans will shop elsewhere or somewhere more convenient.
One older traditional search bar. A 2nd AI chat interface for people who want AI.
I seriously can't believe how simple the solution is, but how much Google refuses to do it. AI search is a totally different product, just make a new tab like Google Images, Google News or Google Finance. Make it Google AI or Google Gemini.
16kB Flash and 1kB SRAM is plenty.
-------
However, the analog capabilities of this thing are nearly trash. One ADC?? No comparators? No OpAmps? Ehhhhhhhh, this is really bare bones by modern standards.
Modern cheap chips differentiate themselves with how many analog components they stuff into one package so that they are easily used in a wide variety of situations.
The only real feature offered here is size. Which is impressive but definitely limiting compared to the rest of TI's MSPM0 (or even MSP430) lineup.
The best this MSPM0-C seems to be able to do, is ADC input -> PWM output. Which is... the bare minimum necessary for mixed signal control.
Resistor+capacitor for anti aliasing on input, resistor+capacitor (smooth the PWM) + transistor for output, and yeah we have a true system with feedback here. But it won't be as reliable or quick as other tools (comparators, OpAmps, or the like)
The L line and G lines have better ADCs, Comparators, better connectivity, and some of them even have on board OpAmps.
Tiny chips have a use of you are tying to build the absolute smallest devices. But hobbyists don't have the equipment to comfortably build things of this size.
Sizing up to VQFN packages and 0603 passives makes more sense for the typical hobbyist.
-----------
As far as how a professional would use this, there are plenty of good uses of chips inside of cables or other kinds of smarts. Like a chip controlling a bunch of LEDs for example, based on voltage measurements elsewhere (there is a 12-bit ADC after all, which means you have rather solid voltage sensing from a few pins).
Basic voltage comparisons + math + crude timer and then a few pins for in/out gets you to a lot of useful projects. Albeit dumb ones. I dunno why anything needs to be this small in particular though.
I'm not seeing any voltage comparators which are really useful for most of those use cases...
Now the MSPM0 series chips have a few OpAmps or a few better ADCs. TI also have proper power metering chips with 16-bit or 24-bit (!!!!!) ADCs for proper power metering.
---------
This is still a 12-bit ADC strapped onto a 8-pin chip. It's useful for something but not that whole list. Power metering? Heh, TI is smoking something or just using AI auto complete.
> Battery charging and management
> Power supplies and power delivery
> • Grid infrastructure
Those really want a comparator (100ns response time) rather than an ADC (10,000ns response time).
You also want zero-crossing detectors on grid infra, or other analog features that simply aren't available at this price point. Just no.
> • Smart metering
Lol very no. 16-bit ADC at a minimum, probably needs 24-bit delta sigma ADC instead (slower ADC but higher precision). TI has a bunch of MSP430 that do the job well.
Alternatively, an OpAmp or other amplifier can zoom into the current sensing circuits and make a 12-bit ADC work, but you need another bigger chip (the OpAmp), at which point you should be using an OpAmp+MCU solution instead.
Plus the "smart" part of metering needs a better communication module than bare bones basic UART. Maybe wireless of some kind or a DSP for custom modem work.
> • Communication modules
Just no. Waayyyyyyy to slow.
> • Personal electronics
> • Connected peripherals and printers
You have bigger, better chips for these.
> • Building security and fire safety
> • Medical and healthcare
> • Lighting
I think these three are legitimate good use cases.
The USSR wanted to make sure they were the ones who built the largest bomb of all time (the Tsar Bomba at 50 MTon). And after that, development on more ferocious weapons has stopped.
Browsing a bit on Microchip's blog, I discovered that the a new line of chips, the AVR LA, was announced this month (June 2026). This surprisingly continues the tradition of 8-bit microcontrollers into the year of 2026.
Looking over the specs sheet, it seems like Microchip is really selling this as a Capacitive Touch controller, with improvements against noise?? (Something about a driven shield + extra boost? Maybe a voltage boost for faster touch sensing?)
I don't fully understand the capacitive touch side, but its something that Microchip has often included in various AVR chips.
The other tidbits I've noticed is how much weaker / cheaper the specs seem to be compared to AVR DD or AVR EB released in the past 5 years. 10-bit ADC (instead of 12-bit on the DD). Fewer comparators. No "special features" (AVR DB had op-amps as part of the package. AVR EA had programmable-gain differential ADC capable of detecting differences of just a dozen microvolts of change....)
AVR LA however? Seemingly all focused on capacitive-touch features. Maybe someone in this topic can talk if this "driven shield" thing is worth all the effort.
That being said: the AVR LA seems to come in at a significantly lower cost, closer to the 50-cents @5k price point.
There's a few misc changes. "TimerA" (which has been present for years) seems to have been completely replaced by "TimerE". Capabilities wise that's fine, TimerE seems better than TimerA in all ways I can think of... but anyone who built code on the older timer might have some code rewriting before they can use this new chip.
Finally, 1.62V is a few hundredths of a volt lower than previous specs (of 1.8V from earlier models). Someone out there probably is excited about ever so slightly better battery lives from this.
-----------
I appreciate the new line of chips, its always nice to see that the AVR team is still active with new things. However, for the hobbyist, I'll say that the AVR DB is probably better overall chip for most hobbyists. AVR DB is still the 8-bit instruction set you know and love, but has a far wider range of sizes and capabilities (including dual-supply, op-amps, more timers, more comparators, better ADC spec, etc. etc.)
AVR LA seems to be specializing for either low-cost, low-voltage, or finally capacitive touch panels. But aside from those other three uses, this feels very much like a "ATTiny", just a far less capable chip compared to AVR DD/DB/EA/EB (which would be like a modern ATMega).
Uhhhh. Sure it is. We stopped nuclear weapons development. At best, rogue countries can catch up to where we are but there's no political will to build even bigger or more powerful bombs anymore. Thats an entire branch of science that we've literally cut off on a worldwide basis.
Science is, and must, be controlled to stay within the realm of useful to the people. The minute it is no longer serving us is the minute we should work on getting rid of it. Fortunately, science isn't a cohesive bathtub where everything must be thrown away with the baby. We can (and do) pick-and-choose what to develop.
I mean, buildings and hardened targets do care. But ships can just move and the missile will miss.
But yes, defenses vs Drones and Cruise missiles is more than sufficient vs Iran. And navigating narrow waters with higher degrees of mobility is better than our other ships.
What is new is Irans willingness to use them. Which skyrocketed after a few missiles assassinated their supreme leader.
No matter how much military we put into the strait, Iran was just going to blow up UAE and Qatari refineries.
And despite all the madman theory of our current President, there's just too many bribes in those countries who have the attention of the President. So Total War is unacceptable to USA leaders.
-------
That being said: I'm still amused that the ship we needed in this fight was the long cancelled Littoral Combat Ship
Which is still useful if you can prove a problem is in NC. It's just not quite as strong as people make it out to be.
The House is a hilarious mix of terrible Gerrymandering (ex: Florida assuming that Latinos will vote Republican) or defeated Gerrymandering (Texas gerrymandering effect being defeated in court).
All in all, it's a hilarious self own where Republicans couldn't even be trusted to Gerrymander correctly and may have made the House a worse situation for themselves. So if anything, the Gerrymandering is seemingly leading to Democrat advantage because of how incompetent Republicans have been.
So as I said before: we are looking at House blowout for Democrats and even a surprise 50/50 odds for Senate. That is a huge change of that happens.
Your invocation of corruption/Gerrymandering doesn't mean anything if you actually look at what has happened. It only matters if Republicans Gerrymandered correctly.
Then the Trump Administration came in (likely after donations from Google, and other tech bros) and suddenly that Google case was dropped.
Regulating against awful behaviors was happening under Biden, and no longer is happening under Trump. It's about as simple as night and day if you are paying attention.
What changes now is that historically, tech firms were largely apolitical. Today they are hard right support, so Democrats weren't used to memes or lack of free viral marketing (etc etc.)
Today, Democrats are finally waking up to the fact that they are being suppressed by both national media and tech media (Twitter and Facebook) and have begun gaining alternative means of getting their messaging out.
Things have gotten worse, but that causes the strategies to shift and the overall political fight to grow stronger.
Senate is statewide so it's innately immune to Gerrymandering. Like.... Do you even know what that word means?
This is important because Anthropic and OpenAI (and likely future companies) are planning to continue to take this money from VTI / VTSAX fund holders.
I'm under no assumption that SpaceX is unique here. This is now "the new normal" moving forward, unless a new law is introduced (or the rules otherwise change).
VTI (Vanguard Total ETF) is $600B + VTSAX (Vanguard Total Index Admiral) is $489B. QQQ is another $480B AUM.
So we're easily talking about $Trillion++ for funds that are described by this post. (VTI, VTSA, QQQ alone). With dozens and dozens of more funds that are participating in this whole "passively buy stocks" scheme being discussed.
VTI arguably because they must play along, being a "total market fund". Nasdaq changed their rules and then force bought 3x more than usual.
These are NOT small funds. Rather significant parts of the US Economy (VTI, QQQ, among many other ETFs and derivatives) were force-bought in last week, and we are now all in for the ride to watch that portion of our money disappear.
--------
S&P500 is an important player who managed to avoid SPCX (1 year rule + requirements of profitability, of which SPCX isn't profitable yet so... lots of disqualifications here). So we have to give credit where credit is due.
VTI / Vanguard Total Market is a grey case. I once believed in it but now with SPCX and other abusers (Anthropic and OpenAI) clearly aiming to specifically screw VTI's strategy over, it might be finally time to call it quits and leave VTI behind. I don't know for what though.
Nasdaq doubling down and specifically changing their rules to help this whole crap is the ultimate level of disgusting. VTI is just being consistent with their strategy. Nasdaq is making special exceptions explicitly for these cases.
Tutors are not necessary. But this is a humanity / language arts problem. Communication with humans is literally the goal. Conversing with other similarly ranked beginners (with a moderator or teacher guiding the group) is among the best practice you can do.
Again, the point of public discussion with other humans is to find all the little mistakes and misunderstandings. The things that other humans find difficult with your pronunciation, rhythm or accent.
Tutors are simply the most expensive version of this available of live, one on one practice. It's a bad overall $$ value compared to class discussion or clubs but the human in the loop is perhaps the most important thing here for training yourself.
Again: it's the LLMs overly generous acceptance that I find problematic. When I talk in German at A2 level, other humans easily point out when I'm doing things in "English order" or other similar mistakes (which makes the sentence harder to understand in German).
But with an LLM, the LLM just understands broken German in English order or whenever I leave out a separatable verb or whatever. It autocorrects too much.
Most of the German discord group is mostly annoyed at how awful LLM explanations and hallucinations are.
Communication and teaching is something that teachers and tutors are far better at than LLMs. In fact, most seem to agree that the 30EUR or $40 textbooks with well organized listening/speaking tests are leagues better than any LLM subscription. And it probably will take you months+ of daily work to go up a language level. (a2 to B1 or B1 to B2), if not a year++ if you are more casual at learning...
At best, LLMs are a tool for browsing the free web for other resources.
As a teacher they have several flaws:
1. They understand your broken grammar and work with you -- bad. Real native speakers will struggle with bad grammar and pronunciation. You the human need to feel this constantly so that you know where to improve. Feeling the instinctive disgust from the other human is part of what helps us know what to practice.
2. They fail at coursework. A textbook puts you on the proper course, already graded to the level you are on and with exhaustive layouts of the subjects you are expected to know.
3. They understand your broken grammar and converse with you without a full ability to explain why it's wrong. Yes, this really is big enough to mention twice.
----------
I'd say LLMs are a reasonable tool for maybe finding additional grammar resources (ex: another source explaining N-declension, or other subjects you know you are struggling with). But as a general guide??
They don't know what you don't know. You still fall into the beginner trap of spinning in circles. If you already know what to search for, LLMs can accelerate the process but in my experience all the already available cheap textbooks are better sources of exercises and graded listening/reading material.
-------
IE: if you want to get A2 graded German reading and listening, get something like Hueber Lesehefte / Sicherheit ist nur wins Carsten Tsara blickt nicht durch for like 10 EUR (.mp3 read along and .PDF).
Or the myriad of other graded readers already available.
Your test is simple. Just read and understand the book. It's A2 after all, if you believe yourself to be at A2 level (or are aiming to achieve A2), then just read A2 stuff constantly.
It's just the usual LLM stuff. Beginners get wow'd by the chat bot interface but ignore all the issues with learning with LLMs. And beginners also find the well trod path of textbooks and study to be boring. But at the end of the day, the textbook is a simple and consistently useful tool, while LLMs arent.
All of this is basically the full of shitty planning. Further continuation of the war will likely only fracture the IRGC and make further peace deals even more unreachable.
Now if we actually had competent leadership who could make wartime decisions that actually lead to our benefit, the calculus would be different. But as it stands now, further war only strengthens the worst of the worst actors in Iran.
---------
Like seriously, it's like people didn't study WW2 and our decision to keep the Japanese Emperor alive/protected (why our nukes and firebombings were so far away from the Emperor). We used to think ahead and figure out the correct people worth negotiating with as part of our wartime strategy.