That's just the cost of utilizing the dating service known as college. Being able to spend recklessly and manage other people to do one's busywork is what appeals within that dating market. Nobody wants to date a broke loser, so to speak.
Maybe you don't technically need four years to find the "love of your life", but many people like to play the field, as they say, which takes time. People are, after all, there for four years because, on average, they want to be.
Whenever I see someone quoting economic statistics I look them up on FRED and zoom out a little. Usually I close the article at that point. The "Ongoing collusion" table in this article is interesting, though. Capitalism breaks down without competition.
When prices rise and wages don't keep pace, individual solvency forces substitution of inferior goods. CPI's methodology then updates the basket to reflect the forced substitution and pretend it was voluntary. You think you are looking at rising real wages, but due to the basket methodology you are actually looking at the individual solvency constraint in a mirror. The economy could reduce people to eating bug burgers in homeless encampments and that line would still go up through the entire process.
Repeat the exercise using a real deflator and the results are different. Ideally, this would mean constructing a basket of things you (or the people in question) want to buy. In practice, nobody has time for that so people just use an asset with a reputation for holding value (gold) or a proxy for their most important aspiration (housing). These both have problems, but the problems are not nearly as bad as the circular logic in the CPI.
Agree that measuring inflation is complex and expensive. It also glosses over devices getting new features and networks improving latency and bandwidth.
Unfortunately alternative measures are worse. Gold increased in price 37% in the last year. What does that tell me about the price of groceries? Housing prices depend on government committees approving what can be built where. We have no idea what a free market would produce.
No, CPI does not gloss over improvements in quality. CPI gets diligent hedonic adjustments. Good news is always welcome, it seems. By contrast, BLS usually doesn't even try to track enshittification. This compounds, and that's another reason why CPI is a poor measure. Not to mention the recent statistical fuckery with trimmed means or the less recent fuckery around excluding energy costs during bouts of energy inflation. How much proof do you need that CPI is a political quantity not a measure of the economy?
Hypothetical YIMBY parallel universes have no place in an inflation discussion. If NIMBYs inflated the price of houses and you want a house, it's inflation that is relevant to you, end of story. As for the volatility of gold, yeah, that disqualifies gold from being a good measure on short timescales. Fortunately, we have a good short-timescale inflation measure: CPI. But gold compounds correctly whereas CPI does not. In the short run, volatility is everything, in the long run, compounding is everything. Don't use gold to figure out how much inflation happened last year and don't use CPI to figure out how much inflation happened last decade.
Reminder: I presented gold and housing as dirty hacks. The correct methodology is to construct a basket of things you care about and track them. If the dirty hacks have dirty hack problems, well, such is the nature of dirty hacks.
Housing and gold, to be used when the timescale is more than a few years. CPI is fine if the timescale is under a few years, because that is not enough time for its sins to compound.
You posted an incorrect interpretation of the FRED real wages line. All I owe the discussion is an explanation of why I believe your interpretation was incorrect. Which I gave. The fact that I went above and beyond to provide and defend both a theoretical alternative and practical alternatives is a nice extra. Your complaints that the practical alternatives aren't simultaneously perfect and easy is praise by faint damnation. Perfect and easy aren't the ways of this world. No, they only have to be better than CPI on long timescales to be good alternatives -- but the CPI methodology places that bar very low and both gold and housing clear it easily.
My analysis was "Doesn't look so dire to me?" and I don't think the point you raised leads to a different conclusion.
I can tell you know the topic well, but we're at the same place CPI criticism always come back to: "It's flawed, but the alternatives are worse, don't exist, or in some way not practical."
Exactly the opposite: CPI is worse than the alternatives at long timescales, despite their flaws.
I know you really really want to make this about the flaws in the alternatives, but I have spent many sentences dealing with the flaws in those in good faith while you have yet to reciprocate at all with any in defense of CPI, either in comparative or absolute terms. Because you know that's a losing game. CPI is cooked, the FRED real wages graph is cooked, your conclusion from it is cooked, and you know it's cooked.
You seem stuck on the idea that real wages have declined within the last 10 years. Is this vibes or real data? Because nit picking data to justify vibes and dropping the thought there is dumb.
Gold is a market where price moves for reasons unrelated to price level. It was stagnant for decades because rich country consumers value it less and now it's on a run as the US weaponizes the monetary system. It's just not useful to represent inflation over the last 10 years.
Housing is even worse. Unlike dollars and gold, every house is unique and in a different location. It's one of the least useful parts of CPI for the same reason it's a worse deflator on it's own.
I use CPI because it's useful, it exists, and it has less problems than the alternatives on a 10 year chart. We're not comparing to a basket of hand sewn shoes and a horse from the 1820s.
You still haven't engaged with a single one of my CPI critiques. 100% offense, 0% defense, you are not operating in good faith. If you can't defend CPI, maybe you should stop making grand proclamations based on it.
I don't think commonly used inflation is helpful for many people, instead asset price inflation is more interesting and reflects people's feeling more accurate.
If you use asset price inflation, in the last 2 decades, most people's real income consistently dropped.
I think we 80% agree: people should use a measure of inflation that includes more asset inflation than CPI. The American Dream was not to Owner's Imputed Rent a house. However, there's a balance to be struck with reasonable expectations, and "I want to own an increasing fraction of the world economy" isn't a reasonable expectation for everyone to have, but that's what jumping to a wholesale asset inflation measure would get you. The way I pick a midpoint is by fixing the asset: one lump of gold, or one house, are reasonable to use as reference points. But not "one billionth of the revenue from the search monopoly" or "one billionth of all newly mined gold" because those things might be expected to grow without robbing me of anything I care about.
> By sending an email that repels all but the most gullible the scammer gets the most promising marks to self-select, and tilts the true to false positive ratio in his favor.
There's less need to identify the most gullible when AI drops the incremental cost of responding closer to zero.
While he was still living on his own, my grandfather struggled with people sending him scam letters in the mail. He sent a meaningful amount of money to obvious scams despite being a shrewd businessman all his life. AI will take these cons to another level.
Is there any way to help the elderly navigate this shift?
"Is there any way to help the elderly navigate this shift?"
As it stands right now, no. Luckily, at least for now, I'm tech savvy enough to navigate it. But the 100, mostly immigrant, elderly residents of my building will not be so lucky.
Two months ago I learned, during a water outage on a Saturday morning, that my building's new assistant manager was an AI bot from EliseAI. What was an actual emergency, it thought, was something that could be dealt with the following week. It told me that management would respond on Monday, which is completely unacceptable during an emergency water outage. I had no water for 3 days & on Monday the actual human manager waffled on why it wasn't immediately fixed & thought it was cute (he giggled & got excited) that I discovered that the assistant manager was now an AI bot.
And just yesterday, I got an unsolicited text from an unknown number that looks exactly like a phishing scam attempt, except that it was sent by the assistant manager AI bot. The text said that I had $130 in unpaid rent & had links to remedy it. WHAT?!! First, it's illegal in CA to give a tenant notice that way. Second, it looked exactly like a phishing scam attempt. I obviously didn't click the links, but I did parse them & found that they belonged to EliseAI. This also happened a month after the human manager gave out a notice about back rent that I had already asked about. He told me I had $0.00 debt on the books. So why the AI bot phishing attempt?
I perused EliseAI's privacy policy and there is no way that I will be knowingly feeding any PII into it.
I haven't tracked down the manager yet to give him a piece of my mind. But in what world is this kind of activity OK? How are non-tech savvy folks going to survive these things? Especially the large non-English speaking elderly population in my building. As far as I can see (I live in SF & am surrounded by it), we're doomed.
Encourage them to move their money to where it's at least more difficult to access.
Recently I discovered that my mam – a middle-class ex-civil-servant, not rich, but not poor – had £++ sitting in the account that was linked to the Visa card she carries around all day. Why? Just because she hadn't moved it elsewhere. Terrifying, obviously.
All she needed was my encouragement. I think it's in Premium Bonds now: not scam-proof, but at least not trivially scammable.
My elderly father uses a financial planner/wealth management firm.
Their returns are not spectacular and they charge fees but overall they are net positive on returns.
My sister and I were discussing if this was a great idea for him and she pointed out:
"Really, he needs someone he can call to send a check somewhere or send a digital payment and he just calls them and they make it happen."
Ideally, there would be a more generic service that would do this for the elderly. It would be the financial version of those "senior phones" that have big buttons and a "help" button that connects you to a person.
The advice I gave to my folks -- hard ignore any contact attempts not initiated by them. Dont read, dont answer the phone except from recognized numbers. I think thats the only way. No legitimate businesses call or text people out of the blue these days.
I dont believe in AI voice assistants picking up phones and filtering scammers. Those tbings will fight a losing game considering how the mainstream LLMs are built.
It would be great if not for the fact that at least over here in the UK your doctor or anyone from the NHS will always call you from a restricted number. So especially if you're elderly and have regular check ups or anything medical going on, the advice to "only answer known numbers" falls apart. And in my experience NHS receptionists have zero patience for the whole "tell me why you're calling" automated screens too, they just disconnect straight away.
It gets a bit more complicated than that. I don't even regularly need things from my doctor, but I have a referral to a specialist right now, and the waiting time is around 16 months to be contacted by them. At that time they will randomly call me from a restricted number. So I must have all call screening and blocking off for probably 2 years in order to accommodate this. And I can't time it because in the past I was referred to a different one and they contacted me in only 4 months when they said 12+.
Had similar situation here in the US -- though with a more tolerable delay -- with a private health care provider.
The fix of course is an enforced requirement from higher authority that health care providers communicate only by means that prove identity of the source -- never calling from an unverifiable number.
This doesn’t work for older folks. They have medical providers constantly randomly reaching out. Typically medication changes or reminding them it’s time to come do labs or whatever.
That and trying to walk my parents through turning this sort of thing on already was a 3 hour long process. That they will in no way remember even with a checklist.
There is unlikely going to be a technical way to solve this social problem, as is usually the case.
Easier said than done.
One of most popular scam here is a scammer using deepfake/ai calling victim saying there was serious accident to one of victim's relative and they need money for hospital (that's how it works here, money for proper caring). The victim would be panic and send the money.
Usually, the scammer will think of a scenario that give the victim very limited time to stop and think, or put them in the state that fear takes over their mind
In places where we have that the scam is just money for transport, utilities, or whatever. Something has happened which means they cannot access their money and make the usual payment.
We actually had an experience with a bank fraud team recently and they relentlessly ask over and over "is the recipient in financial difficulty?". They're not passing judgement on them but trying to work out if you've been convinced that they are and you need to send them money [urgently] to help.
My aunt was recently hit by this scam, they used a deep fake of a teenager and she really believed it was my cousin. The lie was that he was stuck in Montreal and needed money for the next flight home.
This happened in Canada. I don’t think having socialized healthcare makes as much of a difference as one would hope.
The problem is that many scammers now groom your brain into trusting them before they even mention money. This is a psychological thing where you can’t necessarily just tell yourself “I’m not going to fall for that.” There are reports from professional scambaiters who, after engaging with a phone scammer for a while, found themselves inexplicably clinging to the belief the scheme was real.
This is possibly due to how our brains are wired to trust an already-established narrative and due to the effect that repeated interactions have on trust. When the scammer eventually steers the topic towards money, phenomena such as the Semmelweis reflex [0] may have already kicked in, which can make it super difficult to re-frame the person as a scammer.
It sounds extremely obvious and infinitely practical.
How often are you (or your parents) legitimately asked for money via email / internet? I mean, my answer is: never. My parents' answer is: never. My parents-in-law answer is: never.
The scam is usually a phone call from a loved one in jail or the hospital. It immediately creates a sense of panic and fully explains the random phone number. The voice doesn't have to sound like you either.
For voice scams this is critical. I set up duress codes with all of my non technical and elderly family and close friends last year.
It should be a simple, pink-elephant-type nonsense question and answer that you rehearse a few times in person and make sure they never share with others. “Q: What’s your favorite leprechaun? A: Rutabaga” or something. Even cognitively impaired folks can, with a few rehearsals over a day or so, retain that easily, I’ve found.
Reminding them to practice it when people reach out pretending to be me takes a bit more rehearsal, but I’ve found that it sticks reliably even then.
The scams have to be culturally relevant to be effective, which is why they set out wide bets looking to see what sticks with a fairly small target audience.
An addendum is don't give money when you're called by a "legitimate contact", call them back first. The scammer is remote so they can pretend to be the "legitimate contact" but not so easy to block the real legitimate contact from being contacted for validation.
I had a similar problem with phone scams for my elder parents. Installed Asterisk with a Raspberry on their land line to allow only whitelisted numbers and the problem was solved. I guess someone has to step in at some point. Problem is not every parent has a tech literate child and the industry doesn’t seem to bother to offer decent solutions. All call blocker hardware that I found didn’t allow whitelist functionality. I guess there’s a business opportunity here easily, build hardware/software solutions to protect the elderly.
Honestly, not really. The problem is that often the scams are symptoms of a larger problem.
The elderly still want autonomy, and the scams adapt and change faster than education can get most of the elderly caught up to the risks. That's entirely removed from the technical aspects, which are a whole other level of education on risk. Scams always will move faster than defenders and efforts to block them.
All of which are divorced from the real problem, which is isolation of the elderly. The elderly often jump in to these scams simply because they just want someone to chat with. Many romance scammers will report that all they do is strike up a conversation with people, and quite often that's all it takes to get in deeper on an emotional level than the victim's entire family. The family treats this like a money problem, when it's an emotional gap the scammer is exploiting.
I had a neighbor once who was so poor her gas got turned off. Blind due to diabetes. Still managed to send half her meager social security check to a televangelist, because the televangelist was on TV in her face all day, and her family visited once a month. The elderly were prime candidates for 1-900 psychic numbers for decades because it was someone to talk to. I imagine your grandfather wanted to feel like he was helping someone, and scam letters gave him a perfect way to convince himself he was doing exactly that.
As an alternative to actually talking for them and being near them, maaaaaybe. Though, at that point you're introducing trust to something that attackers might learn to emulate enough to make it part of the scam one day.
The solution is that in the West, many Boomers are isolated from their families because many of us move for work or college. Get them some friends, a community, a local shuffleboard group or something so they're not so desperate for any connection that they answer every phone call and email with desperation.
No, there is no way to help them. But the coming generations of elderly will be more resistant, as they've grown up in an all-scam society.
You could stand in real life in front of your parents or grandparents while they're talking on the phone to a scammer pretending to be you, and they would rather believe the scammer than believe you.
The current generation of elderly have great disdain and distrust for their children and grandchildren, and the highest respect and deference towards any perceived authority, and especially if they are contacting them through mail or phone. It's just the hyena mindset: Kiss ass of those you perceive to be higher in rank than you, kick down on those you perceive to be lower in rank than you.
You could be the chairman of the Federal Reserve, and your elderly relatives would rather believe a scammer calling and pretending to be their banker than believe you.
I hope you’re right. Anecdotally based on the significant number of young, digital native people I know who are getting successfully scammed, I’m not sure.
I dont think so. The LLMs are trained to play it safe and avoid responsibility for hard decisions. No commercial model will say "No, I wont connect your call, I dont think you are bleeding to death at this moment, I believe you are a scammer". Nope.
neutralize largest threat surface, work your way down.
First is email. Have their login. Then can-spam everything from nigerian princes. Usually if they have been pig butchered in some way before, they are "hot targets". SO they will continue to receive emails until the algo has been retrained to reject the massive amount of junk they likely get. Set up tons of filters as well.
The 2nd threat surface are landlines, since they usually have been around for a while and the number has been passed around. Force VM-only for unknown callers. Then have VM transcribed and sent over to you. Scammers will not usually leave VM so its an autofilter. Anything important, you can relay. With LLMs this becomes considerably easier.
Open-weight and OSS are wildly different and the article makes a poor comparison.
What's the incentive for the Chinese labs to continue releasing weights 5 years from now? It's not a stable equilibrium and cannot last.
- The lab spending large sums on research and training does not get the inference revenue to fund those efforts.
- Unlike OSS where a single volunteer can keep a project going, training costs run into the $billions.
- OSS is often a two way street where features and integrations are built that the original author benefits from. Open weight models are largely a one way street because the marginal benefit is so much less than training costs.
In the short term, it means Chinese labs can attract talent and, I suspect, funding from their gov. Similar to every other industry the CCP subsidized to take over.
> - Unlike OSS where a single volunteer can keep a project going, training costs run into the $billions.
Just some thought: Wouldn't it make sense to build some kind of volunteer computing project to train the next-generation LLM by volunteers, similar to the BOINC [1] projects or Folding@home [2]?
N.B.: BOINC was particularly famous for SETI@home (completed), Einstein@Home, Rosetta@home and PrimeGrid.
I still remember the time when Einstein@Home was in its heyday, and many people who loved putting together fast PCs contributed sometimes even for the reason of showing off in the statistics [3].
> I’ll be impressed if somebody can make that work considering the vastly larger compute required.
I think you underestimate the computational ressources that the mentioned (and similar-kinded) scientific projects needed. Also consider how much computational ressources people invested into cryptocurrency mining.
No, I think the reasons are different:
- Many companies that train AI model use training data which must not be distributed for copyright reasons (and using it is a legal gray zone)x.
- Also consider that the amount of training data is insane. Scientific projects (and cryptocurrency mining, too) have the property that typically the amount of data (storage requirements) is small (or at least the computation can be partitioned so that each sub-task needs little data), but the required computing ressources are insane.
- AI companies consider a huge part of their training data as their "secret sauce" (they often even paid lots of money to generate it, for example by paying world-renowned experts for writing an answer for some important question).
Thus: Yes, the required computation ressources are huge, but this is a problem for which I consider it to be plausible that it can be solved. The real problems are in my opinion different.
China has no end of money to support these companies. The reason this equilibrium is unstable is that the autonomous agentic coding aspect of the models has been so successfully improved that it will soon be a threat to China state security.
What's the incentive for the Chinese labs to continue releasing weights 5 years from now?
In the short term it attracts talent and builds brand, but they make little money on inference to support research and training costs. Tin foil hat thinking: it also pulls inference revenue away from Antropic/OpenAI and a financial crises at those organizations improves the relative position of Chinese labs.
Is there a reason to think open-weight models are a stable outcome? Open source software provides a collaboration framework for engineers from many companies to work together. Model weights are mostly a one way street.
> What's the incentive for the Chinese labs to continue releasing weights 5 years from now?
The ongoing money from their government. The absolute collapse of OpenAI/Anthropic. US economy getting fucked because their bright financiers decided that going all in on the funny text generation machine was a good idea. Continuing to take a dump on US imposed copyright. The gigantic amount of soft power being the ones releasing "open" models grants. The fact that the ongoing AI war has made China LESS reliant on the US and are now producing their own GPUs, RAM and have massively caught up to nvidia. The lists is endless, and half the points more or less boil down to "taking a dump on the US is morally right", and the other half that massive government programs lead to giant leaps that benefit your society more than a dozen VCs on coke ever could.
> Tin foil hat warning: it also pulls inference revenue away from the Antropic/OpenAI and a financial crises at those organizations improves the relative position of Chinese labs
How is that a "tin foil hat" argument? That's how competition works. You want to make your competitors stumble and fall.
> What's the incentive for the Chinese labs to continue releasing weights 5 years from now?
Engineers love to build things (just look at the whole world of open source), and building an AI model is one of the most exciting things to work on. It only takes a few million dollars of funding to produce an AI model. People spend millions on artwork and paintings for fun and prestige. I can easily see why a billionare or government would want to have their own AI model, but is not interested in the business of selling it, so they release it for free. Combined with engineering talent that loves a meaty problem like AI, I see absolutely no reason why open source AI will not continue to thrive, and not just in China.
Maybe more coop programs and less class time.
reply