It's like "using dice to deny claims". Yeah... I guess? But they are the ones performing the actions. What they base those actions on is besides the point.
The point is that denying claims, particularly denying >90% of claims as the first generation of "AI" did (a D20 model), or denying claims requested by doctors without a Very Good Reason, is tantamount to refusing medical treatment to people who have already paid to be covered for medical treatment, and may stand to die if their medical conditions are not treated, as a result of second-guessing the doctor.
It also dramatically reduces the productivity of doctors, who are being filibustered by a system that forces ever-increasing amounts of pre-approvals, charting requirements and appeals on them. If the median doctor can suddenly only perform half as much work... people also die as a result.
It's not behavior that _deserves social protection_, is the point. Allowing companies to participate in social murder & fraud if they claim an AI told them to internally is morally, politically, bureaucratically, and economically abhorrent. Tear it all down and salt the earth with the bankrupt remains of these companies that had the _audacity_ to shift to second-guessing a majority of medical claims.
> Before health insurers reject claims for medical reasons, company doctors must review them, according to insurance laws and regulations in many states. Medical directors are expected to examine patient records, review coverage policies and use their expertise to decide whether to approve or deny claims, regulators said. This process helps avoid unfair denials.
> But the Cigna review system that blocked van Terheyden’s claim bypasses those steps. Medical directors do not see any patient records or put their medical judgment to use, said former company employees familiar with the system. Instead, a computer does the work. A Cigna algorithm flags mismatches between diagnoses and what the company considers acceptable tests and procedures for those ailments. Company doctors then sign off on the denials in batches, according to interviews with former employees who spoke on condition of anonymity.
> “We literally click and submit,” one former Cigna doctor said. “It takes all of 10 seconds to do 50 at a time.”
The problem is not the dice. It's the denials.
RealPage is a service used by landlords to fix prices and reduce competition. It doesn't actually matter that it uses AI to achieve this, they may as well have used in person conversations or a weekly newsletter, what matters is that it's all price collusion. They aren't "using AI to...", they are knowingly and proudly fixing prices to reduce price competition, and AI (or some kind of algorithm at least) may be used somewhere in the data analysis tools.
If online posts and discussion over the last few days, it most definitely does not have social protection.
Though they aren't claiming that it's okay because AI does it. It's that it's profitable and makes it even more so because now they can fire the people whose jobs it was to deny claims.
You can get away with most things in business so long as a pension or retirement fund gets to wet its beak. And that's what private insurance with no public alternative does.
What forces the insurance company to give sound rejections? There are numerous categories of care where you would not expect the individual to survive long enough to bring a lawsuit, or the individuals in question would not have the financial means.
Even if a lawsuit is brought, what obliges the insurance company to be reasonable? The ability to deny medication that I am hearing about certainly are not in my insurance contract.
The contract does say things like "standard of care", which is why they need an actual MD to reject a claim. The AI (or whatever) can help but in the end there is a doctor claiming to read the file and say you didn't need whatever it is.
You can bring a suit, and often the threat alone will work. They're counting on you to give up rather than fight. Or be unable to afford it because most lawyers won't take something like that on contingency.
They're content to just reject the bill or refuse to pay.
They'd love it if you sued the doctor for malpractice on the grounds of recommending something that the insurer's doctor rejected.
I don't think you can use the insurer's doctor who rejected your claim. You can try to get their board certification revoked, and in the wake of the murder people are talking about doing that more. But it's not in itself going to get you your meds.
I don’t follow how the insurance companies doctor escapes liability in this case. If they are interfering with a patients treatment plan, doesn’t that qualify as practicing medicine?
Otherwise the insurance company could easily staff itself with less scrupulous poorly qualified doctors and say “your job is to reject claims”. Why would that doctor have any motivation or requirement to fulfill the duties of a “doctor”?
It pays pretty well and you don't have to see any patients. And as they say, do you know what they call the one who graduates last in med school? "Doctor"
The problem with suing the doctor is that you need a fairly high standard of evidence to prove malpractice. If they have any way to say "this is not utterly unreasonable", they can claim it was your doctor's opinion versus theirs.
And the insurance company has many lawyers looking for ways to let them say that. It's why they have AI rather than just rejecting claims at random.
Though it appears they often do that as well. But the burden is still on you to prove it.
No, they are using collusion to raise rents--AI is simply the latest attempt to whitewash the collusion. Collusion is already illegal. And had been investigated and are currently being prosecuted.
Those prosecutions are now going to magically disappear.
If governments what to make a
simple but effective regulatory change that will improve competitiveness of rent pricing then simply mandate all lease agreements must have their rate and terms published publicly.
Honestly if that happened we’d probably see the end of rent controlled units; people would be pissed to know they’re artificially subsidizing their neighbors 650/month apartment
Personally I’d rather subsidize the units of people in need (which may or may not be people in rent controlled buildings) than spend my money to grow the portfolio of some ultra wealthy corporation. Subsidizing individuals is really not what upsets me.
I’m more concerned how both the vendor and the landlords are dealing with the feedback loop when the inevitable error occurs.
And as far as I’m concerned, I would like the government to mandate data transparency and create the data clearinghouse to facilitate it. Requires APIs be used and open protocols. That’s how we “increase competition and drive down costs for the consumer”. Not just for rents, for insurance, services, and products.
No, reduce barriers to increasing housing supply. You can’t fix the gap between housing starts and population with more barriers to building new housing.
There’s no incentive to increase housing supply in many markets when existing housing stock is algorithmically unavailable for rent to increase rent on what is available. You can watch some apartment complexes list a sudden vacancy as soon as an apartment gets rented.
Basically algorithms would lower prices if there was excess inventory, but landlords figured out they would stop doing so if they only list a subset of apartments. Obviously shortages do occur, but you often see consistent prices even when bursts of new housing come online with a new apartment complex.
So apartment rents are overpriced? Then someone else is incentivised to build a new building, undercut the competition while still making profits since the existing stock was overpriced.
What we should do is mandate that all rent prices be published, and then someone can build a browser extension or website that shows the average and historical prices so renters can quickly spot overpriced rentals.
> Then someone else is incentivised to build a new building, undercut the competition while still making profits since the existing stock was overpriced.
Not when there’s a glut of housing. Undercutting by a new complex results in those same algorithms lowering prices and people aren’t going to move without any incentives, so you end up with a largely vacant complex...
Don’t get me wrong rents aren’t completely uncoupled from reality. But they rarely fall because renting 80% of your stock at historic prices is better than a price war resulting in you still renting 80% of your stock at new lower prices.
> spot overpriced rentals
When prices are manipulated all you’re going to see is the manipulated prices.
Wouldn't fees for un-used housing fix this? Put a cost on with-holding units on the landlords. Otherwise this is an externality where the costs of with-holding is paid by the people who can't find a place to live.
You might be able to make something like that work, but it without care it would discourage new housing and incentivize removing less desirable properties.
A better option might be a tax on unlisted properties, but that’s harder to enforce as renovations after someone leaves isn’t a fixed time period.
Supply is becoming secondary to credit availability and inheritance - those are what drive prices. We can now borrow more than ever, and a decisive advantage in housing is having rich parents.
not when its anti-consumer, these ideas work in free market with multiple suppliers who can upscale and compete.
Real estate is not a free market, due to building regulations, insane municipality fees to pay upfront for housing units (for example, in canada a lot of provinces require housing developers to pay 100k+$ per 2-3bhk apartment upfront before even completing and building housing units, for 4 floor, 3 apartments every floor, thats 100k * 4 * 3 = 1.2 Million$+ in FEES alone before even selling a single unit !)
Plus these systems are being used to force sellers to algorithmically collectively raise rent prices , to max the revenue collected as possible.
This is literally anti competition, anti free market, distributed oligopoly price-fixing like behaviour.
Housing is the largest cartel in Canada, Retail or Media doesnt work in money laundering.
Canadian housing stock is literally used to whitewash illegally earned cash internationally (but especially china), and they influence politicians and municipalities to put in place laws to stifle housing construction.
Canadian housing is used by international criminals as a store of value instead of gold or other stuff, it’s a very popular method of money laundering.
It’s insane, none of this is free market, like some folks are obnoxiously touting in this thread, this is not free market, not pro-competition, not market decided pricing.
Homes should be for living in, just like Cola is used as a consumable, not as gold/bitcoin to store value, it’s being made this way by Politicians stifling the market with insane regulations preventing homes from even being built.
But if your only standard for whether or not something should be done is "does it create monetary value", you would do anything to get that value anyways.
Which is why we have to have regulations to prevent that from happening, because humans will do some pretty screwed-up things to get money.
what else it should be used? also who makes these decisions? should we form a committee to decide what it can and cannot be used for? US only people make decision or we let UN-like entity decide? what is the punishment for using it for unapproved things - death penalty?
> should we form a committee to decide what it can and cannot be used for?
We already do that. Corporate boards.
Every other Western nation has figured out some level of this and people in the US are still asking "who does x?" with a slippery slope argument implied.
Given the murder of UHC's CEO this week, the concept of "social murder" has been doing the rounds [1]:
> When one individual inflicts bodily injury upon another such that death results, we call the deed manslaughter; when the assailant knew in advance that the injury would be fatal, we call his deed murder. But when society places hundreds of proletarians in such a position that they inevitably meet a too early and an unnatural death, one which is quite as much a death by violence as that by the sword or bullet; when it deprives thousands of the necessaries of life, places them under conditions in which they cannot live – forces them, through the strong arm of the law, to remain in such conditions until that death ensues which is the inevitable consequence – knows that these thousands of victims must perish, and yet permits these conditions to remain, its deed is murder just as surely as the deed of the single individual; disguised, malicious murder, murder against which none can defend himself, which does not seem what it is, because no man sees the murderer, because the death of the victim seems a natural one, since the offence is more one of omission than of commission. But murder it remains.
The American private health insurance industry is quite literally built on knowingly and foreseeably killing people (by denying claims) for profit. But we don't tend to view it as murder in the same as we do when someone is gunned down on the street. Yet we should.
Shelter is a basic human need. Lack of shelter will foreseeably kill people. It's no different to private health insurance.
As another commenter put it (correctly), this is really about using AI to collude on pricing If most landlords use the same software system that produces the same results, then that's just collusion.
That's really what AI is for: to detach humans from their responsibility in violent outcomes, be it with colluding to raise rents as is the case here or, say, to target bombs [2].
We're barrelling towards being just like brick kiln workers in Pakistan, India and Bangladesh except half of us are cheering that on.
The thing is, it’s not AI, it’s just price fixing. We are going to see a lot of stuff in the near future where people are doing something illegal and they’re going to say “the AI made me do it!”
This is a bad article. There are two issues that are being conflated: collusion, and algorithmic pricing. On top of this, the author is either ignorant of economic principles, or is deliberately avoiding using economic analysis.
Collusion is already illegal. If RealPage is facilitating collusion, then they and the participating landlords should be prosecuted. RealPage seems to be in a strange and novel business where they're recommending prices that might not be the best for their customers, the landlords. This seems a lot like collusion with extra steps. If they're actually recommending prices that maximize profits for each customer individually then that seems fine.
A quote from the article suggests that RealPage is being used in real estate companies by upper managers to guide and oversee the work of their reports:
> If a property manager disagrees with the price the algorithm suggests and wants to decrease rent rather than increase it, a pricing advisor will “escalate the dispute to the manager’s superior,” prosecutors allege in the suit.
This... is fine as long as the price being recommended is in the best interest of the real estate company rather than RealPage.
Algorithmic pricing is a perfectly normal extension of basic business management. A real estate management company could achieve the same thing that RealPage is doing (aside from collusion) by buying data from other companies and using some fancy math to choose the best rental prices.
The article makes a claim that isn't really substantiated:
> The complaint names specific areas where rents are artificially high.
> In the second quarter of 2020, the average rent in San Diego County was $1,926, reflecting a 26% increase over three years, according to the San Diego Union-Tribune. Rents have since risen even more in the city of San Diego, to $2,336 per month as of November 2024 – up 21% from 2020, according to RentCafe and the Tribune. That’s 50% higher than the national average rent.
This is a non sequitur. How San Diego rents compare to the national average doesn't tell you anything about whether rents are artificially high. San Diego is a desirable place to live; maybe the rents are artificially low and really should be 2x the national average! What would tell you that is whether landlords could make more money by lowering rents. This would only make sense if occupancy rates are low and landlords could fill more empty units by lowering rents. It wouldn't necessarily make sense even if occupancy rates are low, as high prices filter out bad tenants who cause damage and other costs and can't compensate the landlord because they don't have money. In some markets filtering out these bad tenants can be worth it even if units go empty. California has really poor protections for landlords against bad tenants, so I wouldn't be surprised if higher prices and lower occupancy rates was the correct business decision for landlords there.
The problem in this case is not AI, this is all about the data.
RealPage has tons of data about the properties in an area. That data helps to know if your rent is below the market or if the market demand is growing.
While you can apply ML techniques to suggest a rent price, it could also be done with Excel and a few formulas. Their secret sauce is not the algorithm.
Real Estate investors don't trust generative AI, for a good reason: LLMs are good for summarizing offer memorandums and other documents, but their hallucinations -and lack of context- make it a terrible choice for investment recommendations. The hallucination part could be reduced with better data quality, but then the key is again: data.
The more profound implications that the RealPage case entails is how all the data collection and optimization helps companies to drive revenue at the expense of some people's quality of life: the optimization features for landlords go against tenants.
This happens in almost every product: the optimizations in newspaper's clickthru favor sensationalistic articles with low quality. This is not a new problem, the biggest difference is that nowadays we leave a huge data trail that favors these shitty optimizations, and market competition is not going to magically fix it.
59 comments
[ 6.0 ms ] story [ 124 ms ] threadI'm sure they also use paperwork to deny claims.
Would you make that into a headline?
It's like "using dice to deny claims". Yeah... I guess? But they are the ones performing the actions. What they base those actions on is besides the point.
The point is that denying claims, particularly denying >90% of claims as the first generation of "AI" did (a D20 model), or denying claims requested by doctors without a Very Good Reason, is tantamount to refusing medical treatment to people who have already paid to be covered for medical treatment, and may stand to die if their medical conditions are not treated, as a result of second-guessing the doctor.
It also dramatically reduces the productivity of doctors, who are being filibustered by a system that forces ever-increasing amounts of pre-approvals, charting requirements and appeals on them. If the median doctor can suddenly only perform half as much work... people also die as a result.
It's not behavior that _deserves social protection_, is the point. Allowing companies to participate in social murder & fraud if they claim an AI told them to internally is morally, politically, bureaucratically, and economically abhorrent. Tear it all down and salt the earth with the bankrupt remains of these companies that had the _audacity_ to shift to second-guessing a majority of medical claims.
> Before health insurers reject claims for medical reasons, company doctors must review them, according to insurance laws and regulations in many states. Medical directors are expected to examine patient records, review coverage policies and use their expertise to decide whether to approve or deny claims, regulators said. This process helps avoid unfair denials.
> But the Cigna review system that blocked van Terheyden’s claim bypasses those steps. Medical directors do not see any patient records or put their medical judgment to use, said former company employees familiar with the system. Instead, a computer does the work. A Cigna algorithm flags mismatches between diagnoses and what the company considers acceptable tests and procedures for those ailments. Company doctors then sign off on the denials in batches, according to interviews with former employees who spoke on condition of anonymity.
> “We literally click and submit,” one former Cigna doctor said. “It takes all of 10 seconds to do 50 at a time.”
The problem is not the dice. It's the denials.
RealPage is a service used by landlords to fix prices and reduce competition. It doesn't actually matter that it uses AI to achieve this, they may as well have used in person conversations or a weekly newsletter, what matters is that it's all price collusion. They aren't "using AI to...", they are knowingly and proudly fixing prices to reduce price competition, and AI (or some kind of algorithm at least) may be used somewhere in the data analysis tools.
Though they aren't claiming that it's okay because AI does it. It's that it's profitable and makes it even more so because now they can fire the people whose jobs it was to deny claims.
You can get away with most things in business so long as a pension or retirement fund gets to wet its beak. And that's what private insurance with no public alternative does.
(Not expecting the former, but it needs to be explicit as the distinction is kinda important...)
Either the insurers are getting it dangerously wrong, or we've got problems even worse than that.
Even if a lawsuit is brought, what obliges the insurance company to be reasonable? The ability to deny medication that I am hearing about certainly are not in my insurance contract.
You can bring a suit, and often the threat alone will work. They're counting on you to give up rather than fight. Or be unable to afford it because most lawyers won't take something like that on contingency.
They'd love it if you sued the doctor for malpractice on the grounds of recommending something that the insurer's doctor rejected.
I don't think you can use the insurer's doctor who rejected your claim. You can try to get their board certification revoked, and in the wake of the murder people are talking about doing that more. But it's not in itself going to get you your meds.
Otherwise the insurance company could easily staff itself with less scrupulous poorly qualified doctors and say “your job is to reject claims”. Why would that doctor have any motivation or requirement to fulfill the duties of a “doctor”?
The problem with suing the doctor is that you need a fairly high standard of evidence to prove malpractice. If they have any way to say "this is not utterly unreasonable", they can claim it was your doctor's opinion versus theirs.
And the insurance company has many lawyers looking for ways to let them say that. It's why they have AI rather than just rejecting claims at random.
Though it appears they often do that as well. But the burden is still on you to prove it.
Those prosecutions are now going to magically disappear.
Elections have consequences. ¯\_(ツ)_/¯.
We wanted unending and ever-increasing returns to shareholders, because somehow that's supposed to get the rest of us money.
So (not me, the collective) we voted for a guy who promised that.
I know exactly how people benefitting from government's housing programs are paying, because it is written on the website.
I’m more concerned how both the vendor and the landlords are dealing with the feedback loop when the inevitable error occurs.
And as far as I’m concerned, I would like the government to mandate data transparency and create the data clearinghouse to facilitate it. Requires APIs be used and open protocols. That’s how we “increase competition and drive down costs for the consumer”. Not just for rents, for insurance, services, and products.
Basically algorithms would lower prices if there was excess inventory, but landlords figured out they would stop doing so if they only list a subset of apartments. Obviously shortages do occur, but you often see consistent prices even when bursts of new housing come online with a new apartment complex.
What we should do is mandate that all rent prices be published, and then someone can build a browser extension or website that shows the average and historical prices so renters can quickly spot overpriced rentals.
Not when there’s a glut of housing. Undercutting by a new complex results in those same algorithms lowering prices and people aren’t going to move without any incentives, so you end up with a largely vacant complex...
Don’t get me wrong rents aren’t completely uncoupled from reality. But they rarely fall because renting 80% of your stock at historic prices is better than a price war resulting in you still renting 80% of your stock at new lower prices.
> spot overpriced rentals
When prices are manipulated all you’re going to see is the manipulated prices.
A better option might be a tax on unlisted properties, but that’s harder to enforce as renovations after someone leaves isn’t a fixed time period.
Real estate is not a free market, due to building regulations, insane municipality fees to pay upfront for housing units (for example, in canada a lot of provinces require housing developers to pay 100k+$ per 2-3bhk apartment upfront before even completing and building housing units, for 4 floor, 3 apartments every floor, thats 100k * 4 * 3 = 1.2 Million$+ in FEES alone before even selling a single unit !)
Plus these systems are being used to force sellers to algorithmically collectively raise rent prices , to max the revenue collected as possible.
This is literally anti competition, anti free market, distributed oligopoly price-fixing like behaviour.
Canadian housing stock is literally used to whitewash illegally earned cash internationally (but especially china), and they influence politicians and municipalities to put in place laws to stifle housing construction.
Canadian housing is used by international criminals as a store of value instead of gold or other stuff, it’s a very popular method of money laundering.
It’s insane, none of this is free market, like some folks are obnoxiously touting in this thread, this is not free market, not pro-competition, not market decided pricing.
Homes should be for living in, just like Cola is used as a consumable, not as gold/bitcoin to store value, it’s being made this way by Politicians stifling the market with insane regulations preventing homes from even being built.
But if your only standard for whether or not something should be done is "does it create monetary value", you would do anything to get that value anyways.
Which is why we have to have regulations to prevent that from happening, because humans will do some pretty screwed-up things to get money.
We already do that. Corporate boards.
Every other Western nation has figured out some level of this and people in the US are still asking "who does x?" with a slippery slope argument implied.
> When one individual inflicts bodily injury upon another such that death results, we call the deed manslaughter; when the assailant knew in advance that the injury would be fatal, we call his deed murder. But when society places hundreds of proletarians in such a position that they inevitably meet a too early and an unnatural death, one which is quite as much a death by violence as that by the sword or bullet; when it deprives thousands of the necessaries of life, places them under conditions in which they cannot live – forces them, through the strong arm of the law, to remain in such conditions until that death ensues which is the inevitable consequence – knows that these thousands of victims must perish, and yet permits these conditions to remain, its deed is murder just as surely as the deed of the single individual; disguised, malicious murder, murder against which none can defend himself, which does not seem what it is, because no man sees the murderer, because the death of the victim seems a natural one, since the offence is more one of omission than of commission. But murder it remains.
The American private health insurance industry is quite literally built on knowingly and foreseeably killing people (by denying claims) for profit. But we don't tend to view it as murder in the same as we do when someone is gunned down on the street. Yet we should.
Shelter is a basic human need. Lack of shelter will foreseeably kill people. It's no different to private health insurance.
As another commenter put it (correctly), this is really about using AI to collude on pricing If most landlords use the same software system that produces the same results, then that's just collusion.
That's really what AI is for: to detach humans from their responsibility in violent outcomes, be it with colluding to raise rents as is the case here or, say, to target bombs [2].
We're barrelling towards being just like brick kiln workers in Pakistan, India and Bangladesh except half of us are cheering that on.
[1]: https://en.wikipedia.org/wiki/Social_murder
[2]: https://www.vox.com/future-perfect/24151437/ai-israel-gaza-w...
Some earlier discussion on The Markup original, avoid syndicated content:
https://news.ycombinator.com/item?id=42303530
DOJ sues realpage for algorithmic pricing scheme that harms renters
https://news.ycombinator.com/item?id=41330007
Canadian mega landlord using AI 'pricing scheme' as it hikes rents
https://news.ycombinator.com/item?id=41452781
Collusion is already illegal. If RealPage is facilitating collusion, then they and the participating landlords should be prosecuted. RealPage seems to be in a strange and novel business where they're recommending prices that might not be the best for their customers, the landlords. This seems a lot like collusion with extra steps. If they're actually recommending prices that maximize profits for each customer individually then that seems fine.
A quote from the article suggests that RealPage is being used in real estate companies by upper managers to guide and oversee the work of their reports:
> If a property manager disagrees with the price the algorithm suggests and wants to decrease rent rather than increase it, a pricing advisor will “escalate the dispute to the manager’s superior,” prosecutors allege in the suit.
This... is fine as long as the price being recommended is in the best interest of the real estate company rather than RealPage.
Algorithmic pricing is a perfectly normal extension of basic business management. A real estate management company could achieve the same thing that RealPage is doing (aside from collusion) by buying data from other companies and using some fancy math to choose the best rental prices.
The article makes a claim that isn't really substantiated:
> The complaint names specific areas where rents are artificially high.
> In the second quarter of 2020, the average rent in San Diego County was $1,926, reflecting a 26% increase over three years, according to the San Diego Union-Tribune. Rents have since risen even more in the city of San Diego, to $2,336 per month as of November 2024 – up 21% from 2020, according to RentCafe and the Tribune. That’s 50% higher than the national average rent.
This is a non sequitur. How San Diego rents compare to the national average doesn't tell you anything about whether rents are artificially high. San Diego is a desirable place to live; maybe the rents are artificially low and really should be 2x the national average! What would tell you that is whether landlords could make more money by lowering rents. This would only make sense if occupancy rates are low and landlords could fill more empty units by lowering rents. It wouldn't necessarily make sense even if occupancy rates are low, as high prices filter out bad tenants who cause damage and other costs and can't compensate the landlord because they don't have money. In some markets filtering out these bad tenants can be worth it even if units go empty. California has really poor protections for landlords against bad tenants, so I wouldn't be surprised if higher prices and lower occupancy rates was the correct business decision for landlords there.
The problem in this case is not AI, this is all about the data.
RealPage has tons of data about the properties in an area. That data helps to know if your rent is below the market or if the market demand is growing.
While you can apply ML techniques to suggest a rent price, it could also be done with Excel and a few formulas. Their secret sauce is not the algorithm.
Real Estate investors don't trust generative AI, for a good reason: LLMs are good for summarizing offer memorandums and other documents, but their hallucinations -and lack of context- make it a terrible choice for investment recommendations. The hallucination part could be reduced with better data quality, but then the key is again: data.
The more profound implications that the RealPage case entails is how all the data collection and optimization helps companies to drive revenue at the expense of some people's quality of life: the optimization features for landlords go against tenants.
This happens in almost every product: the optimizations in newspaper's clickthru favor sensationalistic articles with low quality. This is not a new problem, the biggest difference is that nowadays we leave a huge data trail that favors these shitty optimizations, and market competition is not going to magically fix it.