Seems like there's a relatively simple solution for this: allow airline tickets to be resold.
Apart from the fact that this restriction massively benefits airlines, I've never understood why it's not allowed. No matter what, you still have to go through security and present your ID there, so the additional security benefit of knowing a traveler's identity at the time of purchase seems pretty marginal.
That only works with concerts because of severely constrained supply. There's only one Taylor Swift, and she can only perform so many times each year. Yes there's probably no shortage of artists in general, but they're not interchangeable (certainly much less so compared to airlines). The air travel market is, therefore, not nearly as constrained.
I dunno, ticket reselling might not be strictly better, but it seems unlikely to be worse and at the very least it would be nice to have the option of selling your ticket if your plans change. As it is, you generally either have to pay extra in advance for that privilege, or just eat the loss. Either way it's great for the airlines, therefore not great for consumers.
I'm imagining a website that functions like a commodity trading website that allows anyone to buy and sell tickets for a given airline/time/location. Airlines could buy back tickets if they wanted.
The Civil Rights Act requires companies not to discriminate against legally protected classes. That includes both disparate treatment and disparate impact.
I can't see how "personalized pricing" would remain unchallenged, given those hurdles, if it truly means pricing fares based on individual characteristics.
It's totally fine to have loyalty programs that reward people who fly Delta frequently with better rates.
But when an ML model is making inferences about a person's willingness to pay based on what data a company has collected about them, that really feels like it's moving into murky waters, even apart from the predatory element.
There's complexity here that could sway a court. A few years ago, we engaged an AI company to identify potentially problematic hospitality customers. Those were required to check in in person at the front desk instead of the app. Sounds like it could fail those tests, but those tests are for humans and this is AI!
the second it leaks that "potentially problematic hospitality customers" disproportionally target a minority you and the AI company will be sued aggressively from all sides.
Maybe their excuse will be that it can't be proven whether or not the deep learning model is discriminating based on this or that, it's just taking in as much data as possible?
I thought the bar was that you had to prove that it doesn't discriminate which a is much higher, I thought this is what drove many settlements.
Personally I would rather limit all price discriminations to a simple predictable and publicly inspectable customizable formulae. It's like the finance guys who exalt the virtues of increased liquidity but that disappears the moment you try to use it. Similar for the extra complexity for price discovery.
Assuming their model is accurate at predicting economic value and assuming that the wage gap between men and women still exists, wouldn't it be relatively straightforward to show men as a group are quoted higher prices?
As a student of political science, I agree with you. However, the current political administration and Supreme Court seem unlikely to be interested in protecting consumers in this way.
You can’t charge black people systemically different prices than white people, for example. Neither better nor worse prices. Proving causality would likely be the hard part, but a systematic survey of disparate pricing could show disparate impact. AI models frequently are invisibly biased on race because of some feature that operates as a proxy measure that the developers don’t recognize. eg something as simple as ZIP code combined with a name is a highly reliable predictor of race, and there are many similar, far more subtle ways bias can creep into a model.
How can anyone allege discrimination when all you have is an AI telling you the price you get? It’s not open source and you don’t get to see data across all races or sexes or whatever to spot a pattern.
I think we just need new regulations and laws. Maybe make income or wealth a protected trait you cannot discriminate on. Or require transparency and reporting on algorithmic pricing. Or higher taxes for businesses that use these tactics.
But that’s years away, if ever. I bet all airlines will do this. I’ll just fly even less
> “make income or wealth a protected trait you cannot discriminate on”
Seems like that would be problematic for progressive tax rates.
Honestly it’s pretty clever. “Wealth is a protected class” would fit right in with the modern right-wing rhetoric, continuing the inversion theme already established with “actually whites are the targets of racism” and “actually Christians are persecuted in the West”, etc.
Time to open source some adversarial AI! De-wealth your data footprint for better prices, leave digital traces that say “I’m dirt poor, but willing to buy if the price is right”.
I think the OSS community might be able to fight this fight.
I don't think that's possible, there's too much data that pairs your real identity with purchases.
Even Delta's own information on your past purchases (recall tickets are nominative) could be enough to size you up.
I don't know who's responsible, but there's clear information sharing related to using a credit card in the US. E-mail address and whether you have Amazon Prime are two examples I've seen (I was never asked for permission to share this information).
Then there's smartphone-based payment methods, by Google and Apple no less. I'm sure that results in no data at all being disseminated.
I don't think it'd be too complicated for a few major retailers, Amazon and airlines to get together and have a good estimate of your income (or at least of your expenses). What are you going to do against that? Pay by cash everywhere?
I dopn't know, could be doable, at least some basics to muddy the water and maybe more:
Use packages like selenium, VPN providers could offer more turn-key w/ advances features as a value add SaaS.
The idea is, you're not blocking the pricing models, you're feeding them a perfectly consistent bottom-tier consumer profile. They don't love you as a customer, but they'll take you, they know you're a reliable buyer for the right price. You just never buy unless it's at the absolute floor. There's be the need for regular cat & mouse tactice, like with VPN & IP blocking those trying to get around geo blackouts.
a) Use headless Selenium automation to simulate a stable consumer footprint: flight searches, cart abandonment, loyalty program logins, price comparisons.
b) Enrichment networks (LiveRamp, Oracle, etc.) pick up your cookies, browsing behavior, email patterns, so give them exactly what they want: budget travel habits, low-income signals, and no upsell interest.
c) Keep a consistent synthetic persona: same email, same billing ZIP, same bargain behavior, every time. Never act like a deal hunter, just a reluctant flyer with no flexibility.
d) Optional: pair it with a residential-looking PO box address ++ curated device fingerprint (e.g., Firefox on Win10, 1366x768). Avoid exit nodes from sketchy VPNs or known scraper IP pools. Prepaid credit cards.
e) You're not avoiding targeting, you're training the model. Over time, it learns: this user only converts at the floor. That's what it shows you.
Too much effort? Bundle it: Again, imagine a VPN-style service that quietly builds your fake profile over weeks using open-source tools. Privacy meets price manipulation.
Relax, the ticket isn’t $4,999 because you’re broke, it’s pre‑success tax on your inevitable successful exit. They do see your YCombinator IP after all. Or...open the site in Lynx on an old ThinkPad. The fare will drop by half.
If you fly a lot it might pay to get a pre-paid Android beater from CVS to use to buy tickets.
But this is our future boring dystopia. We're not going to get the post-scarcity future from Star Trek but the techno-feudal hellscape of cyberpunk novels. Hooray technology!
As someone who always checks both Uber and Lyft, and always keeps both apps open during the ride so the loser knows I went with the cheaper alternative[0], I'm guessing I won't be flying Delta anymore.
[0]no proof, but presumably one or both of their analytics teams have figured out this signal, given its prevalence in the population
For better or worse I think Delta will be happy to lose you as a customer. They are not a low-cost carrier, and their current business strategy is focused on charging higher fares for slightly better service than the competitors. As far as they're concerned, price-sensitive consumers who are looking for the cheapest fare can go fly on Frontier or whatever.
I travel around the world a lot, and Delta was by far one of the the worst airlines I have ever taken. I doubt they 'charge extra for slightly better service'.
Readit News
Delta moves to eliminate set prices, use AI to set your personal ticket price - https://news.ycombinator.com/item?id=44596355 - July 2025 (91 comments)
Apart from the fact that this restriction massively benefits airlines, I've never understood why it's not allowed. No matter what, you still have to go through security and present your ID there, so the additional security benefit of knowing a traveler's identity at the time of purchase seems pretty marginal.
I dunno, ticket reselling might not be strictly better, but it seems unlikely to be worse and at the very least it would be nice to have the option of selling your ticket if your plans change. As it is, you generally either have to pay extra in advance for that privilege, or just eat the loss. Either way it's great for the airlines, therefore not great for consumers.
I can't see how "personalized pricing" would remain unchallenged, given those hurdles, if it truly means pricing fares based on individual characteristics.
It's totally fine to have loyalty programs that reward people who fly Delta frequently with better rates.
But when an ML model is making inferences about a person's willingness to pay based on what data a company has collected about them, that really feels like it's moving into murky waters, even apart from the predatory element.
Deleted Comment
Personally I would rather limit all price discriminations to a simple predictable and publicly inspectable customizable formulae. It's like the finance guys who exalt the virtues of increased liquidity but that disappears the moment you try to use it. Similar for the extra complexity for price discovery.
Deleted Comment
What specific protected class is "harmed" by personalized pricing?
And how would you prove that the model discriminated against them based on a protected element?
Deleted Comment
I think we just need new regulations and laws. Maybe make income or wealth a protected trait you cannot discriminate on. Or require transparency and reporting on algorithmic pricing. Or higher taxes for businesses that use these tactics.
But that’s years away, if ever. I bet all airlines will do this. I’ll just fly even less
Seems like that would be problematic for progressive tax rates.
Honestly it’s pretty clever. “Wealth is a protected class” would fit right in with the modern right-wing rhetoric, continuing the inversion theme already established with “actually whites are the targets of racism” and “actually Christians are persecuted in the West”, etc.
I think the OSS community might be able to fight this fight.
Even Delta's own information on your past purchases (recall tickets are nominative) could be enough to size you up.
I don't know who's responsible, but there's clear information sharing related to using a credit card in the US. E-mail address and whether you have Amazon Prime are two examples I've seen (I was never asked for permission to share this information).
Then there's smartphone-based payment methods, by Google and Apple no less. I'm sure that results in no data at all being disseminated.
I don't think it'd be too complicated for a few major retailers, Amazon and airlines to get together and have a good estimate of your income (or at least of your expenses). What are you going to do against that? Pay by cash everywhere?
Use packages like selenium, VPN providers could offer more turn-key w/ advances features as a value add SaaS.
The idea is, you're not blocking the pricing models, you're feeding them a perfectly consistent bottom-tier consumer profile. They don't love you as a customer, but they'll take you, they know you're a reliable buyer for the right price. You just never buy unless it's at the absolute floor. There's be the need for regular cat & mouse tactice, like with VPN & IP blocking those trying to get around geo blackouts.
a) Use headless Selenium automation to simulate a stable consumer footprint: flight searches, cart abandonment, loyalty program logins, price comparisons.
b) Enrichment networks (LiveRamp, Oracle, etc.) pick up your cookies, browsing behavior, email patterns, so give them exactly what they want: budget travel habits, low-income signals, and no upsell interest.
c) Keep a consistent synthetic persona: same email, same billing ZIP, same bargain behavior, every time. Never act like a deal hunter, just a reluctant flyer with no flexibility.
d) Optional: pair it with a residential-looking PO box address ++ curated device fingerprint (e.g., Firefox on Win10, 1366x768). Avoid exit nodes from sketchy VPNs or known scraper IP pools. Prepaid credit cards.
e) You're not avoiding targeting, you're training the model. Over time, it learns: this user only converts at the floor. That's what it shows you.
Too much effort? Bundle it: Again, imagine a VPN-style service that quietly builds your fake profile over weeks using open-source tools. Privacy meets price manipulation.
https://www.theguardian.com/world/2019/dec/17/vladimir-putin...
But this is our future boring dystopia. We're not going to get the post-scarcity future from Star Trek but the techno-feudal hellscape of cyberpunk novels. Hooray technology!
What can possibly justify this being legal? It seems ripe for abuse.
If airlines are allowed to do this, I don't think there'll be a thing we can do, seeing as tickets are nominative.
[0]no proof, but presumably one or both of their analytics teams have figured out this signal, given its prevalence in the population