Really, sensitive government data projects like this one should never be outsourced, if only for national security reasons.
Right. Like the government builds its own F-15's, right? Really, different organizations are good at different things. The FBI is an investigation organization, not software development. Their expertise is (or should be) investigation and they should outsource those things that are not in the area of expertise. Just like I don't build my own house, I pay someone to do that that knows more about it than I do.
What amazes me is that projects like this can "fail." While it's not within the FBI's area of expertise to develop a computer system, it's not rocket science for those in the field. Heck, give me half that--$85 million--and I'll develop the friggin' system myself.
Finally the majority of people in this country do see through their bullshit, and short of another 9/11, there is no way the republicans can stop it.
And, of course, if there is another 9/11 then the Republicans either caused it or allowed it to happen precisely to avoid that from happening, right? I'm not defending anything the administration has done, but the B.S. is just as thick on the other side of the aisle.
I got a chuckle out of that. Of course, the Mennonite was wrong. His point wasn't proven, he was just right in that case but that doesn't prove he was right in all cases. Now, had the guy replied, "Yes!" the the Mennonite's assertion would have been disproven... but, as-is, nothing was proven.:)
Absolutes such as "never" almost never (hehehe) can be proven, but the opposite can be proven by just a single example.
It's also absurd if the license is actually worded as it is worded in TFA: "the program and its derivative work will neither be modified or executed to harm any human being nor through inaction permit any human being to be harmed."
Well, damn, the execution of that program represents a threat to my business plan and my income. I was harmed! Therefore you can't use the program to do anything that harms me financially. Likewise, if it is used in a city bus and that bus runs me over and your program allowed it to happen through inaction, I was harmed, so you can't do that!
I assume the actual wording in the real license is better thought out. As worded above, it's just silly.
I agree with the second paragraph, but not the first. Terrorists have already bombed trains in Spain and UK. If people stopped flying in the US and started taking the train, trains would become the target. Crowded bars have been targeted (Bali anyone?). And so have subways (Japan?).
The point being that terrorists will target any place there is a concentration of people, and that has included all of the above. Not all of those have been targeted in the US yet, but it's not coherent to believe that if trains suddenly became a popular transportation medium in the US that the terrorists would not target them.
I also agree there is more hype than necessary. Yes, maybe a dozen planes were targeted and the threat was real. That doesn't mean the "security" precautions are a logical course of action. But it is illogical to think that trains are somehow off the terrorist's radar and are somehow automatically safe and we wouldn't need the same level of security as we have in airports if trains carried as much traffic as planes do.
However, all too often, the first sign a pilot has strayed into restricted airspace is when a blackhawk helicopter pops down next to them, or they get buzzed by a fighter jet. Radio problems are a recurring theme in the encounters- military aircraft with semi-working civilian-band radios, or military pilots not knowing what frequencies the pilot is on/should be on.) You can't really lean out the window and say "hey, officer, what's the problem?", and GA pilots are faced with a terrible conundrum- clearly someone is pissed, but what to do? Change flightpath, possibly becoming more of a threat? Keep going straight, inadvertently continuing towards whatever everyone is hot and bothered about, and get shot down once they cross some 'line in the sand'? Nevermind that when you've got a guy with a very big machinegun trained on you, flying the plane suddenly becomes the least of your worries, and that's VERY dangerous...
I'm a private pilot. Haven't run into any Blackhawks or fighter jets, but haven't busted any restricted airspace, either. If you're flying, you damn well better know where you are. And before you fly, you should sit down and figure out where you're going to fly and be aware of anything of interest in your proximity. If that's too much to ask of you, please don't take off.
I for one don't plan on doing much flying within this continent with this level of nonsense. Luckily, I'm rather centrally located in Denver so pretty much everything west of Chicago is a potential 1-day (albeit long) drive. But I'd rather spending a day in a car driving through this beautiful country than in an airport with a bunch of idiots that have a problem with my tube of toothpaste. I honestly expect I'll do a lot of driving now instead of flying.
On the other hand, I'm a private pilot so I'm going to seriously look into just flying myself. Not cheaper, but still more fun.
Finally, I'd note that in the article header that it is irrelevant that we don't have a useful nationwide rail system. If we did, the exact same absurd procedures would eventually be implemented there. The only way to avoid them is to not go to places where lots of people make attractive targets. That means driving or flying yourself.
The fact that it's difficult or inconvenient to train the filter doesn't mean that step can be avoided; if it weren't necessary, no-one would do it, even (or especially) those that receive a lot of email. I suggest that any Bayesian filter be assisted by keyword filters during the training phase. But, regardless, there is no way to skip the training process. That is what makes it work, eventually.
Bayesian filters don't learn what is spam. They learn what isn't spam. And that's very user-specific.
If you start using a Bayesian filter before it has time to get a sufficient corpus, it's not going to have trouble detecting the spam even if virtually no-one has identified spam. It's going to have trouble detecting the ham. You're going to find you have a bunch of false positives. As such, having a lot of people reporting spam isn't going to help a group Bayesian filter because there is no problem detecting the spam, it's the ham that's hard to detect.
At the very most, you might be able to get away with a few people that do essentially the same job and talking to essentially the same people sharing a corpus (though I question even that). But if you even go as far out as having developers and accountants sharing a corpus, forget it, not a good idea.
Like I said, I think they are probably targetting a specific kind of filter. Perhaps in this case it would be an organizational Bayesian dictionary. Having run a Baysian system for a group (2,000 users), I can tell you that it is VERY resource intensive to maintain individual dictionaries. I'd say it is MORE resource intensive than maintaining the Email boxes themselves. It is very tempting to use group dictionaries.
Certainly. But anyone that advocates using Bayesian statistics on anything other than an individual level does not have an understanding of Bayesian stats. It should never be done. It's next to useless. If an anti-spam provider suggests a solution that is effectively pooling statistics for multiple users, eject that provider because he either doesn't know what he's talking about or doesn't care about your spam problem.
Statistical-based spam filtering must be done on an INDIVIDUAL basis. Always. No exceptions.
It strikes me that one way to combat botnet spam, might be to write a script that would extract the sending IP from identified spam, and add it to a blacklist
This is the default result of any decent Bayesian filter. Bayesian filters should be looking at the headers, too, so the source IP (and all kinds of other related goodies, such as the route) are going to be gathered and judged by Bayesian just like anything else. No reason to have a special process to blacklist specific IPs. Bayesian is going to accomplish the same thing for you, and in a more statistically sound way.
Yep, every decent Bayesian filter must keep track of three things: 1) # of times seen in good email, 2) # of times seen in spam, 3) Last date seen. At some point, terms that haven't been seen should be purged. What that "cutoff date" is will depend on how much mail and spam you receive. If you don't have much volume, you should keep it longer.
Actually, on second thought, #3 shouldn't be "last date seen." It should be "how many messages have been received since the last time this term was seen."
Any decent Bayesian filter uses headers (not just the "To:" line) as a source of tokens. Many times, whether a message is spam or valid is based entirely on what's in the headers which we normally never even look at. What is in the message body actually becomes irrelevant. Of course, spammers try to "pad" or include bogus/useless headers. Just like those spammers that try to "poison" the stats in the content, such spammers have no understanding of how Bayesian filters actually work and so they don't realize that, at-best, such attempts don't help them and, at worst, actually increases the chances of them being detected as spam.
The point is that telling your bayesian filter that lots of random English text is spam will cause it to generate more false positives, which will render it a liability rather than an asset.
That is a popular, but incorrect, belief. The vast majority of English words in a typical user's Bayesian stats are "neutral." For example, "The" is going to be neutral because it appears in both spam and good email. So the word "the" simply isn't used to determine whether or not a message is spam. Now if you are talking to a programmer, the word "compiler" might be a 91% probability of being a valid message. So you might think that by sending the word "compiler" in a poison mail (which is reported as spam) is going to make his spam filter more prone to false positives, right? Wrong...
Perhaps you succeed at making the "compiler" term drop to 87%... or 60%... or even 30%. Doesn't matter. The word will no longer be 91%, but at 91% it is doubtful that it was being used to judge a message as valid anyway.
Bayesian doesn't look at all words in a message, it looks at the 10 (or 15 or 20, whatever) "most interesting" words. That is, it looks at those words that are furthest away from 50%. So if you talk to some guy named Skywalker, that might be a 99.95% indication of a good message. Likewise, maybe be always sends messages from "point4.city.someprovider.com" so all the sudden "point4" becomes a high indication of a good message, etc. And so on.
The long and short of it is that the chances of any of the terms that are in a "poison" message being terms that your Bayesian filter was actually using to determine "good" email are remote. And even if they get lucky, there'll be plenty of other "good" tokens that will use to correctly flag your good email as good.
If you have software that allows it, try checking out the terms your Bayesian filter uses as "good" terms. I.e., that have spam probabilities below 1%. If you take a look at that list you'll realize just how improbable it is for poison spammers to guess even one of those terms; and to truly poison you, they'd have to guess probably hundreds of them. Unless you regularly discuss Tom Sawyer, sending you an excerpt from Tom Sawyer is only going to increase the probability of it being detected as spam.
No, "poisoning" Bayesian stats is something attempted by those that don't under Bayesian stats. But I enjoy the fact that they're wasting their time, so more power to them!
think that is the point. They want to either poison those words so you get more false positives or they want to push other REAL spam related words out of the "this is spam" dictionaries. Maybe both. If these messages had some common theme, they would all get blocked and would have no net effect. They need you to click "this is spam" to poison your filters. Question is, does it work?
Answer is: No, it won't. At least not with Bayesian. The only way to mess up a Bayesian filter is if they can send you messages that are heavy in words/terms that often appear in your good email. And that's going to vary from user to user. Unless you're sending me the exact words that I use in my daily emails, adding a plethora of other words is not going to make my filter any less accurate or create more false positives. It will either let my filter recognize your "poison" as spam itself or, at worst, be neutral.
My Bayesian filter, among other things, considers an excessive number of infrequently/never used terms as a characteristic that is itself subject to Bayesian classification. So while the "poison words" have no statistical effect on my filter, the fact that a bunch of unusual words are found in a message is going to increase the chance that my filter correctly recognize the message as spam.
My spam was constantly growing through about December of last year. This year, it seems to have leveled off. Sure, I'm still getting just under 20,000 per month which sucks, but I see almost none of them and according to my spam stats, the spam has leveled off. Hopefully this is the plateau before it falls.:)
I still want to know: Who are the idiots who BUY spammed products???
I'm receiving spam that's been getting through my Bayesian filter lately, but I have no idea why. It includes an embedded image so it should be recognized almost instantly as spam. The entire "mail" is in the image. But if you look at the source, they also include a text and an html version with random words that are obviously an attempt to use words that someone might be whitelisting. But I'm not whitelisting them. These messages have been getting through with just slightly under my spam cutoff %. I *think* the Bayesian filter is learning and I think the scores of those that get through are getting ever-closer to the cutoff point; heck, maybe the filter is catching most of them and the few that get through are just the lucky ones. But the fact that there is an embedded image and it's not being immediately flagged as spam is curious since an embedded image is going to a very spammy aspect of the message. At this point, I'm collecting these spams that get through to take a closer look at them.
I'm all for a printer using wireless to receive data, even if only for the cost savings on stupid printer cables, which have always been an appalling racket.
USB cables are no more economical. Granted, you're talking about a wireless printer; but I recently had to buy a USB-A to USB-B cable. $30+!? WTF! I couldn't help but think, "Ah, the day before 'progress' when we just used economical (relatively speaking) $10 serial cables" and didn't have to worry about device drivers for it.:)
Not only that, my Bluetooth mouse (which I loved in theory) would just get erratic. I don't know if there's some kind of interference where I live or the fact that the Bluetooth transmitter/receiver was plugged into the USB port on the back of my laptop and the mouse was off to the right side and so sort of hadto transmit through a little bit of computer. Dunno... But I just got sick of having to 1) Change the batteries every couple of weeks. 2) The erratic behavior of the mouse such that pointing and clicking the right thing on the screen was often impossible or, at least, frustrating.
So I just went back to my 'ol infrared cabled mouse. Works like a charm, not spending on batteries, and I can actually point and click what I want to.
This is Slashdot. They don't let anything get in the way of their leftist agenda, especially not little things like facts or reality. Heck, even when an obvious doctoring job on an image to make things look worse than they are comes out, they still try to pin it on the GOP (see the nutjob a few posts earlier that claimed this was really a right-wing effort to discredit the media, as if the media needed any help).
I'm wondering... did this old guy really deploy the device to annoy the neighbors or to annoy mosquitos? If the guy can't hear it, do you really think he even knows other people can? What, did he hop on to how-old-folks-can-annoy-young-neighbors.com and realize that the mosquito device will emit noise that will bother everyone but that he is too deaf to hear?
Did you think of maybe knocking on his door and politely mentioning that the device makes constant noise? Maybe he doesn't even know it!
So, what, are you advocating the government instead spend money on "busy work"? Roads and bridges, right? Jobs are created by business, not by the government. All government can do is get out of the way as much as possible.
There is plenty of work in the U.S. Maybe not quite as much as during the bubble of the late 90's, but hey, it was a bubble and that sucker popped. People got used to insane salaries and earnings and now that things are reasonable and sustainable, they cry because it doesn't match the bubble. Welcome to reality!
For one aspect, as someone mentions in another thread, AOL has been aggressive about keeping spam under control.
Yeah, by throwing out the good indiscriminantly with the bad. When people contact me about a notification or registration or confirmation email that they didn't receive, it inevitably is an AOL account. I don't even bother to try to fix "their" problem if there's no money involved. This last week, though, someone wrote me three emails (which I replied to three times) saying that he never received his email registration for some software he purchased from me. Each time I explained that AOL was probably throwing out my responses as spam. He eventually gave me a Verizon email address to cc to, which I did. The email was received instantly.
AOL is obsolete and treats its users real email no differently than spam. I strongly recommended that this user abandon AOL and use a real ISP exclusively because his use of AOL was making his inbox unreliable. I know if my ISP was doing that poor a job filtering my spam (by throwing out three good emails one after the other), I'd be cancelling, too.
I did this. I lived in Mexico for the last 10 years for personal reasons and 95% of my business was back in the U.S. Had my U.S. business number forwarding down to me in Mexico. Very few people even realized I wasn't in the U.S. It worked just fine. That said, I recently moved back to the U.S. and I've seen my income increase 300% so I'm not sure that living offshore is really a wash. Of course, I was living fine in Mexico but I certainly can't claim there was no financial impact.
It also depends where you live in Mexico and what you spend money on. Labor costs are lower and some food costs are lower. But cars, electronics, appliances, and just about anything that isn't labor-intensive is more expensive. Which is why so many Mexicans drive up to the U.S. to buy stuff. I'm really enjoying that when I want to buy some obscure electronic thing, I just drive a few miles and pick it up or order it online and it appears at my door in a day or two. That just doesn't happen in Mexico--especially if you live in the "cheap" parts. Where I lived in Mexico, rents and houses were actually more expensive than where I now live in the U.S. (Colorado). So with costs generally being about the same where I lived in Mexico and where I live now, and with the 300% increase in income I've seen, moving back definitely was a good financial decision.
Most people suck, why would I want to know them personally?
Yup, I'd say you belong in the "big city" of concrete where everyone avoids eye contact with everyone else. Hey, folks, if that's the way this guy feels, don't try to convince him otherwise. Let him stay in the city. He sounds like an urban anti-social hermit. Convincing people like this to move to more attractive areas will only make those areas less attractive in the long term.
Slashdot Mirror
Right. Like the government builds its own F-15's, right? Really, different organizations are good at different things. The FBI is an investigation organization, not software development. Their expertise is (or should be) investigation and they should outsource those things that are not in the area of expertise. Just like I don't build my own house, I pay someone to do that that knows more about it than I do.
What amazes me is that projects like this can "fail." While it's not within the FBI's area of expertise to develop a computer system, it's not rocket science for those in the field. Heck, give me half that--$85 million--and I'll develop the friggin' system myself.
And, of course, if there is another 9/11 then the Republicans either caused it or allowed it to happen precisely to avoid that from happening, right? I'm not defending anything the administration has done, but the B.S. is just as thick on the other side of the aisle.
Absolutes such as "never" almost never (hehehe) can be proven, but the opposite can be proven by just a single example.
Well, damn, the execution of that program represents a threat to my business plan and my income. I was harmed! Therefore you can't use the program to do anything that harms me financially. Likewise, if it is used in a city bus and that bus runs me over and your program allowed it to happen through inaction, I was harmed, so you can't do that!
I assume the actual wording in the real license is better thought out. As worded above, it's just silly.
The point being that terrorists will target any place there is a concentration of people, and that has included all of the above. Not all of those have been targeted in the US yet, but it's not coherent to believe that if trains suddenly became a popular transportation medium in the US that the terrorists would not target them.
I also agree there is more hype than necessary. Yes, maybe a dozen planes were targeted and the threat was real. That doesn't mean the "security" precautions are a logical course of action. But it is illogical to think that trains are somehow off the terrorist's radar and are somehow automatically safe and we wouldn't need the same level of security as we have in airports if trains carried as much traffic as planes do.
I'm a private pilot. Haven't run into any Blackhawks or fighter jets, but haven't busted any restricted airspace, either. If you're flying, you damn well better know where you are. And before you fly, you should sit down and figure out where you're going to fly and be aware of anything of interest in your proximity. If that's too much to ask of you, please don't take off.
On the other hand, I'm a private pilot so I'm going to seriously look into just flying myself. Not cheaper, but still more fun.
Finally, I'd note that in the article header that it is irrelevant that we don't have a useful nationwide rail system. If we did, the exact same absurd procedures would eventually be implemented there. The only way to avoid them is to not go to places where lots of people make attractive targets. That means driving or flying yourself.
The fact that it's difficult or inconvenient to train the filter doesn't mean that step can be avoided; if it weren't necessary, no-one would do it, even (or especially) those that receive a lot of email. I suggest that any Bayesian filter be assisted by keyword filters during the training phase. But, regardless, there is no way to skip the training process. That is what makes it work, eventually.
If you start using a Bayesian filter before it has time to get a sufficient corpus, it's not going to have trouble detecting the spam even if virtually no-one has identified spam. It's going to have trouble detecting the ham. You're going to find you have a bunch of false positives. As such, having a lot of people reporting spam isn't going to help a group Bayesian filter because there is no problem detecting the spam, it's the ham that's hard to detect.
At the very most, you might be able to get away with a few people that do essentially the same job and talking to essentially the same people sharing a corpus (though I question even that). But if you even go as far out as having developers and accountants sharing a corpus, forget it, not a good idea.
Certainly. But anyone that advocates using Bayesian statistics on anything other than an individual level does not have an understanding of Bayesian stats. It should never be done. It's next to useless. If an anti-spam provider suggests a solution that is effectively pooling statistics for multiple users, eject that provider because he either doesn't know what he's talking about or doesn't care about your spam problem.
Statistical-based spam filtering must be done on an INDIVIDUAL basis. Always. No exceptions.
This is the default result of any decent Bayesian filter. Bayesian filters should be looking at the headers, too, so the source IP (and all kinds of other related goodies, such as the route) are going to be gathered and judged by Bayesian just like anything else. No reason to have a special process to blacklist specific IPs. Bayesian is going to accomplish the same thing for you, and in a more statistically sound way.
Actually, on second thought, #3 shouldn't be "last date seen." It should be "how many messages have been received since the last time this term was seen."
Any decent Bayesian filter uses headers (not just the "To:" line) as a source of tokens. Many times, whether a message is spam or valid is based entirely on what's in the headers which we normally never even look at. What is in the message body actually becomes irrelevant. Of course, spammers try to "pad" or include bogus/useless headers. Just like those spammers that try to "poison" the stats in the content, such spammers have no understanding of how Bayesian filters actually work and so they don't realize that, at-best, such attempts don't help them and, at worst, actually increases the chances of them being detected as spam.
That is a popular, but incorrect, belief. The vast majority of English words in a typical user's Bayesian stats are "neutral." For example, "The" is going to be neutral because it appears in both spam and good email. So the word "the" simply isn't used to determine whether or not a message is spam. Now if you are talking to a programmer, the word "compiler" might be a 91% probability of being a valid message. So you might think that by sending the word "compiler" in a poison mail (which is reported as spam) is going to make his spam filter more prone to false positives, right? Wrong...
Perhaps you succeed at making the "compiler" term drop to 87%... or 60%... or even 30%. Doesn't matter. The word will no longer be 91%, but at 91% it is doubtful that it was being used to judge a message as valid anyway.
Bayesian doesn't look at all words in a message, it looks at the 10 (or 15 or 20, whatever) "most interesting" words. That is, it looks at those words that are furthest away from 50%. So if you talk to some guy named Skywalker, that might be a 99.95% indication of a good message. Likewise, maybe be always sends messages from "point4.city.someprovider.com" so all the sudden "point4" becomes a high indication of a good message, etc. And so on.
The long and short of it is that the chances of any of the terms that are in a "poison" message being terms that your Bayesian filter was actually using to determine "good" email are remote. And even if they get lucky, there'll be plenty of other "good" tokens that will use to correctly flag your good email as good.
If you have software that allows it, try checking out the terms your Bayesian filter uses as "good" terms. I.e., that have spam probabilities below 1%. If you take a look at that list you'll realize just how improbable it is for poison spammers to guess even one of those terms; and to truly poison you, they'd have to guess probably hundreds of them. Unless you regularly discuss Tom Sawyer, sending you an excerpt from Tom Sawyer is only going to increase the probability of it being detected as spam.
No, "poisoning" Bayesian stats is something attempted by those that don't under Bayesian stats. But I enjoy the fact that they're wasting their time, so more power to them!
Answer is: No, it won't. At least not with Bayesian. The only way to mess up a Bayesian filter is if they can send you messages that are heavy in words/terms that often appear in your good email. And that's going to vary from user to user. Unless you're sending me the exact words that I use in my daily emails, adding a plethora of other words is not going to make my filter any less accurate or create more false positives. It will either let my filter recognize your "poison" as spam itself or, at worst, be neutral.
My Bayesian filter, among other things, considers an excessive number of infrequently/never used terms as a characteristic that is itself subject to Bayesian classification. So while the "poison words" have no statistical effect on my filter, the fact that a bunch of unusual words are found in a message is going to increase the chance that my filter correctly recognize the message as spam.
My spam was constantly growing through about December of last year. This year, it seems to have leveled off. Sure, I'm still getting just under 20,000 per month which sucks, but I see almost none of them and according to my spam stats, the spam has leveled off. Hopefully this is the plateau before it falls. :)
I still want to know: Who are the idiots who BUY spammed products???
USB cables are no more economical. Granted, you're talking about a wireless printer; but I recently had to buy a USB-A to USB-B cable. $30+!? WTF! I couldn't help but think, "Ah, the day before 'progress' when we just used economical (relatively speaking) $10 serial cables" and didn't have to worry about device drivers for it.
So I just went back to my 'ol infrared cabled mouse. Works like a charm, not spending on batteries, and I can actually point and click what I want to.
This is Slashdot. They don't let anything get in the way of their leftist agenda, especially not little things like facts or reality. Heck, even when an obvious doctoring job on an image to make things look worse than they are comes out, they still try to pin it on the GOP (see the nutjob a few posts earlier that claimed this was really a right-wing effort to discredit the media, as if the media needed any help).
Wow, you're about five beers short of a six pack.
Did you think of maybe knocking on his door and politely mentioning that the device makes constant noise? Maybe he doesn't even know it!
There is plenty of work in the U.S. Maybe not quite as much as during the bubble of the late 90's, but hey, it was a bubble and that sucker popped. People got used to insane salaries and earnings and now that things are reasonable and sustainable, they cry because it doesn't match the bubble. Welcome to reality!
Yeah, by throwing out the good indiscriminantly with the bad. When people contact me about a notification or registration or confirmation email that they didn't receive, it inevitably is an AOL account. I don't even bother to try to fix "their" problem if there's no money involved. This last week, though, someone wrote me three emails (which I replied to three times) saying that he never received his email registration for some software he purchased from me. Each time I explained that AOL was probably throwing out my responses as spam. He eventually gave me a Verizon email address to cc to, which I did. The email was received instantly.
AOL is obsolete and treats its users real email no differently than spam. I strongly recommended that this user abandon AOL and use a real ISP exclusively because his use of AOL was making his inbox unreliable. I know if my ISP was doing that poor a job filtering my spam (by throwing out three good emails one after the other), I'd be cancelling, too.
It also depends where you live in Mexico and what you spend money on. Labor costs are lower and some food costs are lower. But cars, electronics, appliances, and just about anything that isn't labor-intensive is more expensive. Which is why so many Mexicans drive up to the U.S. to buy stuff. I'm really enjoying that when I want to buy some obscure electronic thing, I just drive a few miles and pick it up or order it online and it appears at my door in a day or two. That just doesn't happen in Mexico--especially if you live in the "cheap" parts. Where I lived in Mexico, rents and houses were actually more expensive than where I now live in the U.S. (Colorado). So with costs generally being about the same where I lived in Mexico and where I live now, and with the 300% increase in income I've seen, moving back definitely was a good financial decision.
Yup, I'd say you belong in the "big city" of concrete where everyone avoids eye contact with everyone else. Hey, folks, if that's the way this guy feels, don't try to convince him otherwise. Let him stay in the city. He sounds like an urban anti-social hermit. Convincing people like this to move to more attractive areas will only make those areas less attractive in the long term.