Preus personalitzats amb IA: per què el mateix servei pot costar un 30% més
Un experiment troba diferències de fins al 30%, però no totes proven la vigilància. Expliquem els resultats, els límits i què han confirmat FTC i Consumer Reports.
Dues persones poden veure preus diferents per un mateix producte o servei. Els algoritmes ajusten imports per demanda, ubicació, historial, promocions i experiments comercials. Quan també utilitzen dades personals per estimar quant està disposat a pagar cada consumidor, es parla de preus personalitzats o «preus de vigilància».
Chris the Producer intenta manipular aquest sistema durant una setmana. Crea un perfil digital nou, el fa semblar primer poc disposat a comprar i després el situa virtualment en una comunitat rica. Troba diferències de fins al 30%, però el mateix vídeo admet que més de la meitat de les comparacions donaven preus iguals o molt semblants.
1. Preu dinàmic, experiment algorítmic i preu de vigilància
No tota diferència és personalització. Cal distingir:
- preu dinàmic: canvia per hora, demanda, estoc o disponibilitat;
- segmentació geogràfica: depèn de la zona, costos o competència local;
- experiment A/B: usuaris assignats a grups veuen preus diferents;
- promoció: un compte nou, fidel o inactiu rep un descompte;
- preu personalitzat: dades sobre una persona o perfil determinen l’oferta.
Veure dos imports diferents només demostra una diferència. Per atribuir-la a vigilància cal provar que l’algoritme ha utilitzat dades personals i descartar la resta de variables.
Aquest matís és central. El vídeo mostra un fenomen real i opac, però no té accés al model intern d’Uber, DoorDash, Instacart, Target o una agència de viatges.
2. Les primeres comparacions
El creador comença comparant comptes de familiars:
- un producte de Target apareix a 4,99 dòlars per a una persona i 4,59 per a una altra;
- un trajecte programat costa 36,94 en un compte i 52,95 en un altre;
- la mateixa habitació es mostra a 122 dòlars i a 84.
Són diferències grans, però la gravació no documenta totes les condicions: moment exacte, impostos, membresies, cupons, tipus d’habitació, cancel·lació, demanda, distància de recollida o historial promocional.
Serveixen per formular la pregunta, no per respondre quina dada va causar el preu.
3. Un perfil nou no comença necessàriament barat
Per separar-se del seu historial, Chris crea una entitat empresarial, obté un mitjà de pagament i compra un telèfon barat dedicat. L’objectiu és construir una identitat digital sense cookies ni compres prèvies.
En la primera prova, el resultat va en direcció contrària a l’esperada. Una caixa de bolquers a Instacart costa:
- 34,99 dòlars al perfil nou;
- 28,99 al compte habitual.
El perfil acabat de crear paga sis dòlars més, aproximadament un 20%. Això podria ser una cohort experimental, una promoció del compte antic, ubicació, historial o qualsevol altra regla. No prova que la identitat empresarial sigui la causa.
També mostra per què «esborrar cookies» no garanteix el preu mínim. Un perfil sense historial pot perdre descomptes de fidelitat o ser tractat amb més incertesa.
4. Com intenta fabricar una baixa intenció de compra
El creador contracta un actor perquè utilitzi el telèfon durant unes hores. El personatge busca articles barats, visita zones universitàries, abandona compres i només acaba comprant un plàtan de 39 cèntims.
Després, els preus d’Uber que veu aquest perfil són de mitjana un 11% inferiors als del compte de comparació, segons el vídeo.
El comportament és coherent amb la hipòtesi: l’algoritme podria interpretar que l’usuari és molt sensible al preu. Però no hi ha grup de control, repeticions suficients ni accés al sistema. Entre la primera i la segona prova també canvien hora, ubicació i estat de la demanda.
La narració és un experiment periodístic i humorístic, no un assaig causal.
5. Fer-se passar per ric produeix els descomptes més grans
La segona hipòtesi és contraintuïtiva: un consumidor en una zona rica pot rebre un preu inferior perquè té més alternatives i no sembla desesperat.
El creador no pot entrar en una comunitat privada de Minnesota, així que eleva el telèfon amb un dron i el controla remotament des d’una carretera pública. Des d’aquesta ubicació compara:
- un Uber fins al Mall of America: 54,95 dòlars dins la zona i 75,96 fora, un 28% menys;
- una comanda de White Castle a DoorDash: 39,47 i 47,25, un 19% menys.
Els números són reals a les pantalles mostrades, però els serveis no són idèntics si canvia el punt de recollida o lliurament. Distància, conductors disponibles, frontera de zona, temps estimat i tarifa local poden explicar part o tota la diferència.
Per tant, la prova no demostra que l’algoritme premiï la riquesa. Demostra que la ubicació pot coincidir amb una diferència important i que l’usuari no rep una explicació clara.
6. La dada més honesta: més de la meitat de preus coincidien
Abans dels resultats finals, Chris explica que més de la meitat de les proves donaven imports iguals o molt similars i que va seleccionar les diferències més grans per il·lustrar el fenomen.
Aquesta transparència és valuosa, però introdueix biaix de selecció. Si es fan moltes consultes i només es publiquen els extrems, la variació pot semblar més sistemàtica del que és.
Una prova robusta necessitaria:
- centenars de comptes o participants;
- consultes simultànies;
- mateix producte, botiga i ubicació;
- control de cupons, membresia i taxes;
- repeticions durant diversos dies;
- protocol definit abans de veure resultats;
- anàlisi estadística de totes les observacions.
És molt semblant al mètode que Consumer Reports va utilitzar en la seva investigació d’Instacart.
7. Què va demostrar Consumer Reports
Consumer Reports i Groundwork Collaborative van coordinar més de 400 compradors en quatre ciutats. Els participants consultaven simultàniament cistelles idèntiques a les mateixes botigues.
Van trobar:
- diferències en prop de tres quartes parts dels productes;
- fins a cinc preus simultanis per un mateix article;
- variació mitjana d’un 7% en el total de la cistella;
- diferències màximes per article de fins al 23%;
- un impacte anualitzat potencial d’uns 1.200 dòlars per a una família de quatre.
Instacart va confirmar que alguns minoristes feien experiments de preus, però va dir que eren limitats i aleatoris. També va negar utilitzar dades personals o demogràfiques per fixar aquells preus.
Després de la investigació, la plataforma va deixar d’oferir la tecnologia que permetia cobrar imports diferents a compradors diferents pel mateix producte al mateix moment, tot i mantenir proves de descomptes i promocions.
És evidència forta de preus algorítmics diferents. No és, per si sola, prova que cada diferència fos causada per un perfil personal.
8. Què va trobar la Comissió Federal de Comerç
La FTC va requerir informació a intermediaris que ajuden empreses a ajustar preus. L’anàlisi inicial va concloure que aquestes eines poden utilitzar:
- ubicació precisa;
- historial de navegació i compra;
- dades demogràfiques;
- moviments del ratolí;
- productes abandonats al carret;
- canal i moment de compra.
Els intermediaris estudiats treballaven amb almenys 250 clients de sectors com alimentació i moda. La FTC va advertir que es poden personalitzar preus, promocions o l’ordre dels productes.
Com que la informació comercial és confidencial, l’informe públic utilitza exemples hipotètics i dades agregades. No identifica la causa de les diferències concretes del vídeo.
9. Els 8.793 dòlars del titular no són un estalvi auditat
El títol del vídeo afirma haver estalviat 8.793 dòlars. La narració, però, no presenta una cistella real de compres per aquest import ni un càlcul complet que es pugui reproduir.
La xifra sembla dependre d’extrapolar diferències puntuals a una despesa més llarga. Una extrapolació pot explicar el cost potencial, però no equival a diners efectivament estalviats.
El que sí queda documentat són comparacions concretes de pocs dòlars o desenes de dòlars i percentatges de l’11%, 19%, 20%, 28% o 30% en casos seleccionats.
10. Crear una identitat falsa no és una recomanació
El vídeo utilitza una societat, un actor, un telèfon separat i un dron com a recurs narratiu. Reproduir-ho pot vulnerar condicions del servei, normes de pagament, registres empresarials, privacitat o regulació aeronàutica segons jurisdicció i execució.
Tampoc és una defensa pràctica: mantenir una empresa i diversos dispositius costa més que molts dels descomptes observats.
Per comparar preus de manera segura, un consumidor pot:
- consultar diverses botigues o plataformes;
- comparar el total final, no només el preu inicial;
- revisar si hi ha membresia o cupó aplicat;
- comprovar l’oferta iniciant i tancant sessió;
- capturar pantalles simultànies amb una altra persona;
- evitar decidir sota urgència artificial;
- reclamar una diferència no explicada.
No cal fabricar una identitat per documentar que dues ofertes divergeixen.
11. A la Unió Europea hi ha una obligació d’informació
La Directiva (UE) 2019/2161 exigeix informar el consumidor quan el preu s’ha personalitzat a partir d’una decisió automatitzada. La regla s’aplica a preus personalitzats, no al preu dinàmic que canvia per demanda sense perfil individual.
Això no és una prohibició general. És una obligació de transparència i conviu amb el RGPD, les normes de protecció de dades i els drets davant determinades decisions automatitzades.
Si una botiga en línia personalitza el preu però no ho comunica, el consumidor pot conservar proves i consultar l’autoritat de consum o protecció de dades corresponent.
La conclusió
El vídeo il·lustra una realitat incòmoda: el preu digital ja no és necessàriament una etiqueta única i l’usuari sovint no sap quina regla l’ha generat.
Les seves comparacions no permeten afirmar que cada diferència sigui vigilància personal. Algunes poden provenir de demanda, geografia, promocions o experiments aleatoris. Les proves de la FTC i Consumer Reports, però, confirmen que existeix una infraestructura capaç d’utilitzar dades granulars i d’oferir preus diferents.
El problema més sòlid no és que un dron obtingui sempre un descompte. És que el consumidor no pot saber si paga més, per què paga més ni quines dades han decidit el seu preu.
Contrast i context
Fonts consultades
-
01
Chris the Producer How I Tricked Big Tech’s AI Pricing Algorithms to Save $8,793
-
02
Federal Trade Commission Surveillance Pricing Study: initial findings
-
03
Federal Trade Commission Issue Spotlight: The Rise of Surveillance Pricing
- 04
-
05
Consumer Reports Instacart Stops Pricing Tests on Its Platform
- 06
Font de treball
Transcripció amb marques de temps
Consulta la transcripció
-
0:00
, obre el vídeo en una pestanya nova
We're all currently paying different prices for the exact same thing. And this is because of a brand new thing called surveillance prices or valence pricing or valence pricing. It's making life more expensive with some people paying $1,200 more on groceries alone. This is really across all sectors of American shopping. I had never heard of this before, but it's like price tags are dead now. And even though 76% of people think it should be illegal, surveillance pricing is being used by more and more companies. And here's how it works. We kind of already know that for decades, corporations have collected our personal data to sell us targeted ads. But now that same data is being used with an AI algorithm to set the maximum possible price they think you'll pay. And a different maximum price they think I'll pay in a different price for my mom and your mom in a different price for every mom in America. This goes beyond just moms too and I'll prove it. We're gonna buy the exact same product from the exact same store Go to Target and buy spam Okay, how much is it? 499 my price is 459 as Bullshit We're gonna order the exact same ride. Okay scheduled for tomorrow at noon Do it from Target field to their port. Here's my prices $36.94 for the Overseas. Mine's 52.95. That is a 30% difference in price. And for what? No difference. Let's look the exact same hotel on the exact same night. What's the name? La Quinta Hotel Denver. Okay, 120, 250. Oh, $122. Yeah. My price is $84. Wow.
-
1:51
, obre el vídeo en una pestanya nova
Wow! With surveillance pricing or personalized pricing, people are very up in arms like the idea that they might be paying a different price than their new before something. That's Grace Gattie and she works for Consumer Reports. And with her help, I'm going to put the surveillance pricing system to the ultimate test and go to the most extreme lengths possible to gain the system and trick these AI pricing algorithms in order to save as much money as possible. What the f-? Is it possible to avoid personalized pricing? Um, not really. That's when I decided to dedicate my life for the next week to beat surveillance pricing and my first mission would be to get a fresh start. Except that's almost impossible. Your information is scattered across thousands of invisible databases meaning a terrifying accurate profile of you and me and everyone else already exists. So many companies are collecting data about you. Your search history, wait, been reading online, what you clicked on, or hovered over your location data, where you bring your devices, all of this data can be used to build a pretty detailed profile about you and what you're not being willing to pay for individual items. But I don't need to erase myself from the system. I just need to become someone else. And in order to buy things online and get purchasing power from a new credit card, I need a new social security number. But apparently that's not possible and actually kind of super illegal. So instead of becoming someone else, I became some thing else. In the US, a corporate entity gets an employer identification number, which is functionally a social security number for businesses. Plus, the law grants corporations almost the exact same rights as humans. So I use the Wyoming corporate loophole and hired a local proxy to form an LLC on my behalf.
-
3:37
, obre el vídeo en una pestanya nova
This is something you can actually do in a state of Wyoming to legally remove your real identity from the public record. So there is no possible chance that anybody would know that my name, Chris, par, owns this LLC. The guy on the phone, Ray, got the filing set up leaving me with one final task to name the entity. To trick the system, I needed a name that sounded like a guy who technically has cash but who's visible lifestyle and digital footprint guarantees he will never show a high intent to purchase anything. Thank you so much Ray. Appreciate it. I am Frank Reynolds. LLC. Now that I had a new EIA number associated with Frank Reynolds LLC, I could get a credit card, which was actually super easy. And all I needed now was a dedicated device for Frank to make purchases on. But I had to be careful because everything I do in Frank Reynolds' name now has to be carefully executed in order to trick big data. Maybe creating a new identity through the Wyoming corporate loophole isn't for everyone. Fortunately, there's a service that will get your data scrubbed from Data Brokers for you. And that's the sponsor of this video in Cognito. Because our data is scattered everywhere online being bought and sold by mysterious and shadowy data broker sites, you are unknowingly being taken advantage of based on your information, your data that's out there. Yeah, I know it's crazy. It's required by law in a lot of states for data brokers to remove your information if you request it. But it is really hard to request it. Fortunately and Cogny does this work for you, removing you from literally hundreds of data broker sites all on your behalf. I actually tried to do this manually, but after hours and hours of finding where my data even was in navigating this bureaucratic nightmare, I actually couldn't even successfully get myself removed from one data broker company. Go to incogny.com slash Chris the Prodicer today to sign up. It's really cheap. It's really easy. And you can get 60% off if you use the promo code, Chris the Prodicer. It's kind of just inevitable that we're exposed to the data breaches, like targeted fishing scams or identity theft risks. In fact, my information has been breached 10 times, and that's just from one email account. With in Cogniz custom removal feature in the Elmimited Plan and Family and Limited Plan, you can point to any people search site or any website where your personal information is visible,
-
5:58
, obre el vídeo en una pestanya nova
and one of their privacy agents will take care of the rest for you. Protect your privacy by going to incognid.com slash Chris the Producer and use the promo code Chris the Producer for 60% off. Orbits the travel aggregation website started kind of determining that math users might be willing to pay more for hotels than PC users, steering math users towards priceier hotels. And I didn't want priceier things, so I went to Walmart. Excuse me, I'm just looking for the cheapest, possible smartphone. Oh, $20, $19, right? This was the cheapest available smartphone called the track phone, which was good, because I actually wanted this thing to track everything I do and everywhere I go. There's been some great investment journalism by SFGate where they did a controlled experiment. They're looking at a specific hotel room in New York City. And in fact, when they changed their IP address, the price per night for the hotel dropped by like $500 in one instance and $200 in another instance. I can only assume that's because people from the Bay Area generally have a higher income than wherever they change their IP address to. So I headed to the lowest incomes of code in Minnesota to activate my or actually to activate Frank's phone. And for sure, it's right next to where I live, so I guess that's a good thing for this. So I got a part of what's about, right, Exodus place called Hard Times Cafe. It really a biscuits engraved. Anyway, I activated the new phone and opted in to all possible tracking options. We are going to give this phone access to everything that we are doing. Our location, our app usage, our battery, absolutely everything. And then I went onto Instacard and ordered the exact same box of Panper's Diapers. For my toddler, not me, on both my phone and Frank Reynolds' new track phone. This is the track phone 3499 right there. And my phone has the same exact product from the same exact store for $2.899. The price for Frank Reynolds was $6 more. I had used a corporate loophole to create an entirely new identity set up a new bank account and got an incredible credit card just to pay 20% more for the exact same thing I had failed. And maybe a surveillance pricing wasn't even that bad.
-
8:20
, obre el vídeo en una pestanya nova
Maybe the people can benefit from this. But that's when I remembered what Grace told me. Um, I think companies benefit from this. I mean, companies don't tend to pursue a new pricing tactics unless they think they can, it won't increase their profits. So yeah, I didn't fail. We're just getting going here. What did you think? We're only like halfway through the video. Plus this track phone, Frank's track phone, didn't really have any date on it. Its life, Frank's life, had only just begun. It has zero history, one location data point, and no cookies or whatever that's called. I hadn't actually tried to game the system yet, but I had come to realize I'm not Frank Reynolds and I would never be. I'm Chris, the producer, and what I needed was an actor to become Frank Reynolds. So I made a casting call on Craigslist, seeking an improv actor who could portray the person with the lowest intent to buy literally anything. I received hundreds of applicants, but only one person was perfect. Hello, Comrade Trip here. I'm still here in Minneapolis. Comrad was in, so we arranged to meet at a park parking lot. Hey, how's it going? I'll take any direction you have. I needed to give Comrad direction on how to look entirely uninterested in buying things. The only problem was I had absolutely no idea how to do that. So I've been pricing algorithms, actively black boxes, anyone who doesn't work at the company, and we never know how to all of the inputs that are going into a price. But there was one word that kept popping up every article I read. data. And what's maybe most responsible for this surveillance capitalism might just be data brokers. So to find out what data they are collecting and learn more about this pricing system, I reached out to 50 data broken companies. Which by the way, there's like thousands, literally thousands of these companies that we know nothing about in the industry is worth somewhere in the hundreds of billions. Nobody responded. Literally, not one response, not even a no. But I did find a guy named David Dan, who actually invented the phrase Surveillance pricing. Hey David, how's it going? Okay, are you? Surely he would know how this system works. I sent emails to 50 companies.
-
10:35
, obre el vídeo en una pestanya nova
Do you think these companies don't want you to know? Oh, 100% because that's the only way they can get away with it. I guess from what you know, like what data goes into setting someone's price. That includes browser information. It includes your location. It could include your financial history through your bank account, past transactions that you made with the company. Data brokers can take that information, bundle it up, and sell it to other companies. Croker has data profiles that are 63 pages long on average, on individuals. On each person, on each person, yes. So I guess it's just all data. So I told Comrade to just do whatever it takes to become Frank Reynolds, the person with the lowest intent to buy anything to trick the data. I gave him Frank's phone and Frank's credit card, and then I he went out into the world. What he didn't know is that I would follow his every move. I mean, how else was I gonna capture Frank Reynolds tricking big data? Excuse me. Sir guy waiting at the bus stop there in a vest? Okay, thank you. There he is. He's getting on the bus. Where are you going Frank? Frank downloaded the MTA app. But I found out later, he never actually made a purchase on it. Oh, quiet. Oh, thank you. Which is genius, giving signals to the data being collected that he's really not wanting to purchase anything. And then I lost him. But I was prepared for this.
-
12:11
, obre el vídeo en una pestanya nova
Before filming, I put a tracking device on his microphone and his wearing, so I could track his every move. Send McDonald's right now. While Frank was in McDonald's, he downloaded the McDonald's app then Google cheapest item at McDonald's. And then again, he didn't buy anything and just asked for a walk. Just wanted a longer one. Yeah, please. I promise there's a point to all this. Frank basically walked around for a few hours in Dinky Town where all the poor college kids live. Going business to business, asking and searching on his phone for free stuff. I'm wondering how many people you give to a couple of brown rice? Just a couple of brown rice, free. Oh, okay. He did buy a single banana from the Target app and gave it to these guys. Hey did that guy just give you something? Oh did he make me an a-oh cool. That's nice. And he went into great clips. I don't think he can help me at all actually. But anyway after doing all this stuff we went back to the car and I was pretty skeptical this would change the price at all. But then I remembered how terrifyingly accurate big data is. It's very precise. One of my favorite stories around this. story of a woman who started to get web ads, you know, things like diapers and bass and that and just, you know, stuff for a new mother. And she's like, this is ridiculous, I'm not pregnant. Well, she learned like four days later that she was pregnant. And the company knew based on various pieces of data and her spending habits, they knew she was pregnant before she did. Don't like that at all. So yeah, the track phone is picking up Frank's every move, purchase, and abandoned cart. So I always spent 39 cents a day.
-
13:59
, obre el vídeo en una pestanya nova
And we were about to see if everything he just did would change his prices compared to my phone. Uh, we're going to test Uber, right? Yeah, okay, I go from here to your new new depot. Yeah, about $3 cheaper for Uber, right? So that, so that phone is $3 cheaper? Yeah, and then a dollar keeper for waiting save. Why they're all different. On average, Frank's prices on Uber were 11% lower, which was good. But we can do better. And I think I know how to do it. Oh, here, here, here, here. What's this? He had no, never paid for sugar. Frank was probably being perceived by big data as being low income. But that's not necessarily going to get you the lowest possible prices. willingness to pay doesn't necessarily correlate with income staples online was charging higher prices in certain zip codes. And those zip codes were areas that didn't have a hardware store around them. And those correlated pretty well with lower income communities. So lower income communities and those zip codes were getting charged higher prices because they were desperate. There's no other way to get a stapler. So I wanted to do the ultimate price test and trick big data again but this time by becoming super rich. But I guess I've never been rich before so I didn't really know how to do that and I didn't have time to build and scale a business to own equity or whatever. So I figured I just needed to go to a zip code that only rich people can go to. Fortunately, that place exists. I made a video about this a few months ago. It's a private city called North Oaks, which ranks as the wealthiest city in Minnesota. You literally have to live there to go into the city or be an invited guest. It's also the only municipality in America that's not on Google Street View because not even the Google car can go into the city limits. So in that video I made a few months ago, I flew a drone over the city in order to put it under Google Maps. And after publishing that video, I got a legal letter from the city banning me from entering North Oaks property. So I can't just go into North Oaks to test prices. But I had an idea to get around this. Do you think people in North Oak so cheaper than people not in North Oak? Maybe that's how they afford to live there, I hope. There was only one way to find out. So I bought a phone mount and strapped it to the bottom of my drone and headed for the border of North Oak.
-
16:20
, obre el vídeo en una pestanya nova
It's up to you. You're the see man. Actually, first I had to pick up my friend KJ. I needed him for something super important. And then we headed to the border of North Oak. So that is North Oak right there. But technically we are on public property. This is like a, I think it's like a state highway or something right here. So I took Frank's phone onto the drone and had KJ call me. You're in. Okay. Stay going. Then I gave him access to remotely control my phone, or just something you can do if you're like a boomer and need tech help. But in this case, we were going to use it to make a purchase from inside of North Oaks. Because we can't touch the phone once it's like in the air, right? Yeah. All right. We got to get this thing in the air. Okay. And then just like I did to put North Oaks on a Google Maps street view, I flew the drone into unrestricted airspace across the border from what's above North Oaks property. We had now entered into the highest income zip code in Minnesota. Not really the phone is. So then using KJ's phone to remotely access Frank, we ordered an Uber from inside of North Oaks to the Mall of America. It's 54.95. Then using KJ's phone, we ordered the exact same Uber ride to the Mall of America from where we were sitting. What? Oh my god. It's 7596. Okay. That's 21 dollars more. Which adds up to a 28% decrease in price from inside the City of North Oaks. See what I'm saying?
-
17:48
, obre el vídeo en una pestanya nova
Yeah, that's pretty crazy. Okay, what else can we test? And actually before we keep going, I want to be fully transparent about our process. While testing prices during the other two tests, over half of the time, prices were the same or really similar for Frank and myself. I just picked out the biggest price differences to illustrate the point that surveillance pricing is already happening. But in North Oaks, the only test side did had massive price differences. Let's get, let's do DoorDash, what should we order? Why you cancel? A Crave case. That is 30 sliders. Well ordering from Frank was hovering over North Oaks. The price was 39, 47. They're pretty good. And the price for KJ's phone, outside North Oaks, where we were sitting on the highway, was 47.25. What in the? Resulting in a 19% decrease in price from inside the city of North Oaks. And by the way, my hunch is this price difference has nothing to do with whitecastled the best operated company in the world, but because DoorDash is pretty well known for using personal data to price couch. So you're telling me one of the richest communities in the state gets a discount and frank reddled. So we're just thinking about all of this is we have no understanding as to what's going on behind the scenes. There's seemingly no rhyme or reason for the difference in crisis. And that's precisely why it's objectively unfair to us as consumers. But the reality is that you need that government needs the step in the smallest because you're just as an individual consumer you're not going to be able to fight this system alone. And there are introductions of these kinds of bills and I think to one other states. up. I think there's only one thing we can do now. What's that?