Slashdot Mirror
Do Firefox Users Pay More For Car Loans?
RandyOo writes "Someone wrote in to The Consumerist to report an interesting discovery: while shopping online for a car loan, Capital One offered him different rates, depending on the browser he used! Firefox yielded the highest rate at 3.5%, Opera took second place with 3.1%, Safari was only 2.7%, and finally, Google's Chrome browser afforded him the best rate of all: 2.3%!
A commenter on the article claims to have been previously employed by Capital One, and writes: If you model the risk and revenue of applicants, the type of browser shows up as a significant variable. Browsers do predict an account's performance to some degree, and it will affect the rates you will view. It isn't a marketing test. I was still a bit dubious, but at least one of her previous comments backs up her claims to have worked for a credit card company.
Considering the outcry after it was discovered that Amazon was experimenting with variable pricing a few years back, it seems surprising that consumers would be punished (or rewarded), based solely on the browser they happen to be using at the time!"
Who's saying it is? Correlation is really all insurance rates need to be based on.
which is totally what she said
What the interest rate for IE was?
What rate did they offer for IE6?
Surely that must be good indicator for negligent behaviour?
Actually, since the bulk of people still on IE6 are locked-in due to corporate policy, you could just as easily assume they'd be shown the lowest rate, as they're likely to have a stable, corporate job.
Repeat after me "People who blindly yell 'Correlation is not causation' should be slapped with a trout.'
FTA: "I figured it had just gone up since I received the email. I tried to use their little payment calculator but the flash based widget wouldn't work properly in the Firefox Beta so I loaded up Safari to try and funny enough the rate offered was 2.7%. I checked in Chrome and Opera to see if it was maybe just something wrong with the Firefox beta and Chrome's rate was 2.3% while Opera's was 3.1%."
and "Devin installed fresh versions of the browsers in order to make sure the changes didn't result from different cookie settings. It seems those looking for a Capital One loan should apply through Chrome."
There are no other obvious variables. The only crime you can punish this guy for is not repeating the experiment across other computers. You can try it for yourself to see if it holds true for you as well.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
I got a quote for -£240 using Lynx
I work at IHOP you insensitive clod!
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
Correlation may not imply causation, but it certainly does waggle its eyebrows meaningfully in causations direction.
Here is a comment from the article. I have not tested this.
bearymore::: ""Wow! I have a Firefox addon which spoofs the user agent. When I go to Capitalone with the default Firefox as the user agent I get 3.50%. When set the addon to tell Capitalone I'm using Internet Explorer, I get 2.70%. When I switch back to Firefox, I get 3.50% again. Keep in mind, I'm using the same browser and simply opening the site in different tabs""
So either deliberate or incredibly bad coding for a financial site.
The phrase "correlation is not causation" is strongly correlated with stupidity.
Just turn JavaScript off and you'll get the higher 3.50% every time ...
Seems like something's up with their variables; A different cg variable causes a different rate for the same zip code.
This calls for a Firefox add-on or greasemonkey script that will take such pages, request the pages with different user-agents, and compare the resulting pages for differences. If the only difference is a single numerical value, it should be easy enough to catch. And then print it to the user in a neat table or graph.
Must be the default before Javascript calculates it...
Analogies don't equal equalities, they are merely somewhat analogous.
Retaliation for Ad-Block!
http://xkcd.com/552/
Obviously you fail to understand how my superior browser makes me a better human being than you. As such, I will disregard the rest of your post. It is likely just to bore me, or possibly even lead to a headache. When you realize the error of your ways, and are willing to admit that I am right in this and all things, I will once again indulge you.
SJW: Someone who has run out of real oppression, and has to fake it.
Installing fresh browser versions doesn't clear your existing cookies...
Insurance is a game of statistics. Yes, some browsers lie in their user agent string. Yes, sometimes people use other people's browsers. But MOST browsers are what they say they are, MOST people use their own machine pretty regularly, and if the insurance company is really doing this (which I take no position on), then you can bet they have statistics to back up their belief that there's a difference in insurance risk, ON AVERAGE, between users of different browsers.
The insurance company couldn't care less if the correlation holds true for every single instance. They know that it doesn't, in fact. But if it holds true often enough, then they can use that data to offer some people a better price on the insurance, because they're statistically less likely to file a claim.
Next thing you will get a visit from the fraud squad.
Same as the guy which used Lynx to donate to a charity campaign in the UK a few years back.
The "security pros" providing security solutions t the donation site decided that it is being hacked (well, he did play a bit with various URLs to get past some particularly stupid javascript based submit code).
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
"Correlation is not equal to causation" is what an unintelligent person says when they wish to sound intelligent. it's something they once heard they thought was clever, and they think that by aping this simpleminded thought they are adding something to the conversation, when they are just generating useless noise
an intelligent person would actually be looking at the merits or lack thereof of the correlation, and talking about if causation is implied or not
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
I would guess that IE trends higher is the less tech savvy regardless of income or education. It's those with tech knowledge that deviate from it (not in all cases, but the majority).
they could use that information to find people using apple products. and charge such people 9.5%. as long as the car was titanium white, such people would happily purchase at that rate
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
It seems just asking for your age would be a better way of determining... well, your age.
Just Google for it, clearly Google search robots must get the best rates so that more people will follow the links.
Statistics motherfucker, do you understand them?
Blindly stating 'Correlation is not equal to causation' is not evidence against causation.
Technically, i think the W3C standards for HTML 4.0 require a 3.5% interest rate. The other browsers are just providing proprietary rates in a standards breaking battle of oneupmanship. I think HTML 5 attempts to address this issue with the new heavily scriptable InterestCanvas element.
While it may be true that users of different browsers tend to have varying levels of credit worthiness that tracks to what browser they use, this creditworthiness will be more clearly (and accurately) reflected by other criteria. As an example, while people with a particualar credit score range may be more likely to select a particular browser, it seems improbable to me that people who select a particular browser are more or less likely to default on debt than people with the same credit score who select a different browser.
The truth is that all men having power ought to be mistrusted. James Madison
Sure, but in this case it makes perfect sense. Obviously anyone using Firefox is a filthy open source hippy who demands everything for free. If you tried to loan money to a Firefox user he'd probably just spend it all on weed and then claim your demands to pay back the money violate his privacy in some way.
Opera users aren't good people to lend to either, since none of them have any income. They spend all of their time scanning the Internet for stories about browsers (any browser) so they can jump in and extol the virtues of Opera. This leaves no time to hold down a real job.
Safari users are good people to lend to because, since they're using an Apple product, you already know they're accustomed to paying huge premiums for slightly shinier versions of various consumer goods. All you have to do is send the bill in an elegantly designed box and they'll pay it without question every month.
Chrome users are the best because their close relationship with Google shows they've already given up on the whole concept of privacy and will gladly supply you with any information you ask for. Plus, all you have to do is tell them you're not evil and not only will they allow you to do whatever you want no matter how evil it is, they'll actually defend your actions to others!
Clearly browser choice is and should be a significant factor in the lending business.
It is sometimes. For example, they should be charging the Safari users more, as those Mac fanboys will pay anything if it's marketed right.
"He who can destroy a thing, controls a thing." --Paul Atreides, Dune
Used to work there. Capital One has done this for many, many years. They used to send snail mailers with different rates to random households in the same neighborhood. The CEO is one of the early proponents of experimental marketing.
The entire insurance industry is based on the principle of retaliation against customers who cost the company money and rewards for those who dutifully pay more than they claim. Any statistical basis they can use to figure out which customer is which ahead of time... they'll use. The only ones they (usually) won't use are those prohibited by law; instead they look for some other factor that correlates strongly, and use that instead. So if they aren't allowed to use race, they'll use neighborhood... if they aren't allowed to use neighborhood, they'll look for something else. Maybe ISP, or IP address block, or OS, or.... browser. They don't care, as long as there's a good correlation. It's not that they hate Firefox users, any more than they hate people of a particular race or ethnic group or neighborhood or religion or credit score or driving record... they just don't care. Like any "good capitalist" they want to maximize profit.
http://alternatives.rzero.com/
"Correlation is so highly correlated to causation that we're forced to conclude that correlations cause causation."
Still cracks me up. Probably caused by^W^W^WHighly correlated with reading hundreds of threads like this on Slashdot...
I suffer from attention surplus disorder.
Only when used in idiotic contexts like the first post.
In this case, they are looking for patterns, and the actual cause is irrelevant, if they notice patterns that indicate higher risk, regardless of the root cause, they follow them.
Now, there are correlations (violent video game players have a higher incidence of violent activity) that, by themselves, do not imply causation (further tests are needed for causation). When someone tries to use these as an implication of causation, then the phrase "correlation does not imply causation" is quite intelligent.
And, I just implicitly defended Capital One. I feel morally dirty now. Thanks a lot.
Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
I just spoke to my wife, who spent 10 years working for Capital One at their corporate headquarters in Richmond (*). When I mentioned the gist of the story to her, her response was "I believe it", which actually floored me as I am always telling her about various conspiracy stories I see online. She immediately offered up that they use various bits of financial information to determine what % interest rate you would be charged and suggested that the Safari users would be seen as having more money (as Mac owners) and hence a lower risk.
(*) As an aside, if you haven't been to the Capital One campus in Richmond then you are missing out. The selection of restaurants is amazing. They have outdoor basketball and volleyball courts. And they even have a frickin' treehouse where you can go and sit outside with your laptop and do your work via Wi-Fi. The downside is that corporate doesn't instill loyalty as pretty well everyone in Richmond either knows someone who was downsized from Capital One, or is downsized themselves.
The only better work conditions nearby would be CarMax, which for their staff they don't offer fixed length vacations. You get to choose how long you take off each year - they assume that if you are professional enough to do your job that you are professional enough to know when your job is done!
After reading about this, I am completely {outraged}{amused}{indifferent}{turned on} by this practice.
Javascript error. Aborting script.
Yeah, I'd feel dirty if I defended Capital One, too.
They're local here and known by many people as Crapital One for their firing sprees and tendency to make employees disappear. They're also the fastest bank in the nation to sue their own customers. (I sell data to bankruptcy lawyers who keep up with this kind of thing.) They're the last bank I'd go to for a credit card or a loan anyway.
This is indeed true. I've experienced it this past September. However, I think it is not just based on browser, but also based on geographic location. I checked their site from my home PC with Firefox (east coast state), and to be a bit safer, I VPNed in to my office with my work laptop, still using Firefox, to actually fill out the application. To my surprise, the rate was higher (the exit node to the Internet is a north, central US state for the corporation). I called my wife in to the room to show here two computers, both on Capital One's auto loan home page, showing her the two different "as low as" rates on the page. I'm glad I didn't go with them for the auto load - seedy bastards.
I recently noticed the same type of happening when I was shopping for a Verisign SSL cert (clients *sighs* don't get me started on why). When I use Safari or Firefox, the price for a 1-year "Secure Site" SSL cert (with site seal) is $499 - however, when I switch to Chrome or IE8, I get a price of $399. I only had to buy the one, and so wrote it off as a fluke - but I just re-verified that this is still happening for me (tested on both Mac and Windows), given the news on this article.
This smacks me as being seriously wrong - now I have to test all browsers when buying something online, to be sure I'm getting the right price? And no, I'm not going to change my default browser habits, just to get lower pricing...
Worrying works!! 99% of all the stuff I worry about never happens
The phrase "correlation is not causation" is strongly correlated with stupidity.
Yes, but correlation is not ... oh, wait ...
Quidnam Latine loqui modo coepi?
True, but where there is correlation you have to look for causation. Occam's razor -- Can you come up with a better explanation as to why one would get different rates using different browsers? If not, then you have to suspect causation. If so, then you have to find another cause.
Free Martian Whores!
Tests with car loans:
IE 8: 3.10%
FF 3: 2.30%
Chrome: 3.50%
Is what rates do you actually get? Just because they print a rate on their site doesn't mean you may not get a higher or lower rate when you actually get approved for the product, after they've looked at things that actually matter.
I have trouble believing a random poster that claims to have worked there that this actually influences things. Really? Then where is the actuary data on this, and why isn't it something asked on home loans? Anything that has a significant predictive value int terms of loan performance, banks want to know. Given that home loans are some of the most major loans, they check everything they legally can. If browsers really were such a major predictor, they'd ask on your home loan app.
To me this looks like an error. I cannot believe that the browser a person uses is a good enough performance predictor that they'd offer a 1.2% difference in interest rate (which in the context of these loans is a 35% difference in interest). That is like 100 FICO score kind of interest difference, or more. No way the browser someone uses accurately predicts loan performance to the point of being worth it.
Just remember that the rates on the site are advertising rates. Who know show the hell they decide what it should show or why JS is involved, but that it hows it has nothing to do with the rate you'd actually get when you apply. Not only is the rate they show always the "Very best credit," rate but it often has other provisos. For example when I was looking at a home equity line of credit, the bank had a rate listed on their site that, when you dug in to it, only applied if you got a $250,000 or larger line of credit AND took out $100,000 or more AND kept it out for several months. Ok well that isn't unrealistic (sometimes you finance home construction with such a loan) but clearly most peopel aren't getting that rate.
This is just the advertised rate, which is always based on parlor tricks and weird math. Your actual rate is calculated after your application and credit check are reviewed, and neither of those things care what browser you use.
This makes sense if you think about it. The bank wants to advertise a rate that is appropriate to you, but it doesn't know much about you until you apply. So it has to guess based on the very limited information it does have. Otherwise -- and this still happens all the time -- you see an ad that says 'Low low 2% APR!' and then you apply and find out that you qualify for 6% APR and get pissed off. If looking at your USER_AGENT reduces the spread between what they promise and what they get, it makes sense for them to do it. But don't let that get in the way of your populist rage.
Insurance companies don't care about causation. Correlation is all that really matters to them. If people with names beginning with 'A' have more accidents, insurance companies are going to give them higher premiums. The causation doesn't matter.
And not actual rate differences. Remember that nothing they list there is anything but marketing. Your rate isn't your rate until you actually apply. A number on a website is just promotion/information. It isn't a binding offer of any kind. They'd need a credit check at the very least before they'd be willing to make an actual offer.
Now, as you said, the one and only concern with loans is established risk. What factors (that they are legally allowed to consider) increase or decrease risk? All they are interested in is coming up with a probability of default, and then deciding if they want to make the loan and at what rate based on that.
Well browsers just don't seem to play a role. I've heard of no studies, no evidence, that choice in browser indicates anything with regards to handling money. Without such research, it would be a meaningless metric to use.
Also notice that the rate difference is LARGE. Something like that means a significant risk difference. Even if browsers factored in, you really think it'd be that much? You think they'd offer over a percent just based on the browser?
Not hardly. Unless someone can provide some actuary data showing that this has been looked at scientifically, I do not believe it is actually a deciding factor, just some Javascript being squirrely.
Isn't this whole argument rather moot? I mean, OK, the guy goes to websites for various loan providers, and depending upon his browser, he claims that he is shown different interest rates. Big deal. It's not as if the actual loan is going to be based on the interest rate that was quoted in the little "payment calculator" tool, since any lending institution is going to have to do a credit check on anyone before they loan him $30,000.
Or in the case of fields like marketing and insurance, causation is irrelevant if the correlation is strong enough.
If teenagers have a 1% chance of getting into an accident in their first year being insured with the company, but ones with red cars have a 3% rate, you don't have to prove any causal relationship between car color and ability to drive, you just play the numbers and raise rates accordingly. Business decisions don't have to be rational, they just need to be right 51% of the time to stay in business.
As a general principle, no, you don't. As a rule of thumb, you are probably safer assuming that a correlation between two variables is the result of a common cause. E.g., in humans, height over 6'5 is strongly correlated with usage of urinals. At no point should you bother to rule out causation (in either direction) in this case.
A more practical example is the media's constant repetition that healthier people have more sex. Of course, they jump to the sexiest but dumbest possible conclusion, that the sex causes the health. But we don't even have to assume that the health causes the sex; in fact we can intuit that health must be only partially the cause, since many healthy behaviors will increase attractiveness completely aside from their health benefits. (Especially diet and its effect on body shape.)
The general principle is that any one cause will have myriad effects, all of which will be correlated with one another. By default, assume a common cause.
The sad thing is that you have to use your choice of browser to define yourself and make yourself feel superior to others. I don't need to resort to such immature behavior to validate my self worth. I'm a Mac owner.
Alex, I'll take keybindings not used by Emacs for $400....
I worked briefly in car insurance, and one of the many questions asked by one of the top insurer was how long the client usually kept his car. (Note that this was an information asked from brokers about their customers, not direct b2c)
Turns out people who renew their car often take better care of them so as to maximise resale value, and consequently produce less claims. It wasn't a big difference but it was apparently statistically significant.
Netscape users are not good because they are likely so old, that they are libel to die soon anyway causing you to pay out...
Gopher users are not good because they are dead, and the dead don't pay premiums.
Mosaic users...
I mean, there would be no reason to stop at the browser used? The mode of internet access---heck, even IP address---could be factored in as well.
To-do List: Receive telemarketing call during a tornado warning. Check.
People who blindly yell 'Correlation is not causation' should be slapped with a trout.'
Yeah, but where would we get a trout? For me, fishing is NOT correlated with obtaining a trout, it's correlated with getting angry or drunk.
Sounds like a good opportunity for a firefox plugin.
Got Code?
It's Capital One. They are probably just fucking with people.
Why, without your clothes, you're naked, Miss Dudley!
Why should the interest rate depend on a credit check..?
Are you being serious? Have you never secured credit before in your life? ANY lending institution is going to require a credit check prior to lending you any money. You'll notice those advertisements usually have fine print somewhere which indicate "rates as low as" regarding the rate(s) quoted. Your credit score always comes into play, whether or not you realize it. If it didn't, banks would have to offer the same rate to a guy who defaults on all his loans that they offer to a guy who's paid everything on time for years.
Why should the interest rate depend on a credit check..? If I was advertised a certain rate on the website and then they turned around and said "actually, the real interest rate is 5%", I'd tell them where to go.
And they'd tell you where to go.
And you'd have to walk, because you didn't get the car loan.
Repeat after me
after me
after me
after me
after me
after me
after me
after me
after me
Let q be a radix > 1. I am in ur base-q, killing 10 d00ds.
Yeah, I'd feel dirty if I defended Capital One, too.
They're local here and known by many people as Crapital One for their firing sprees and tendency to make employees disappear. They're also the fastest bank in the nation to sue their own customers. (I sell data to bankruptcy lawyers who keep up with this kind of thing.) They're the last bank I'd go to for a credit card or a loan anyway.
Yeah, but you have to balance that against the fact that they have funny vikings in their ads. That counts for a lot when you're choosing a credit card, you know.
On Slashdot, it's stupid and irrelevant more often than not. People just post it because they have read it before and seen it get +5, insightful. You see it posted every single time there's a story about some indicator of something or other. So the correlation is certainly there.
I think there might be a weak causal relationship as well, since so many slashbots seem to just break down and stop thinking when they come across indicators and other correlations, believing the right answer is to mindlessly harp on about the same old meme, and thus spread it -- in the same mindless form. Most people intuitively understand the difference between correlation and causation, but when they are infected with this nasty little meme, they seem to lose the ability to understand what a correlation is.
I don't have any version of IE, but do have a few browsers installed on Ubuntu 10.04 amd64. FWIW, my IP resolves to somewhere in Finland. Here are the rates offered:
Opera 10.63 = 3.50% new cars, 5.09% used
Konqueror 4.4.2 = 3.50% new cars, 5.09% used
Firefox 3.6.12 = 3.10% new cars, 4.49% used
Chromium 9.0.568 = 2.70% new cars, 4.09% used
Epiphany browser 2.30.2 = 2.70% new cars, 4.09% used
Hitting refresh did not change the rates offered, even if all cookies were deleted.
Those who can make you believe absurdities can make you commit atrocities. - Voltaire