Honest question, is that sarcasm? In particular, the comment about HD video. I've been using MythTV as an open source DVR/media center for many years. When Sandy Bridge came out, it was in theory the perfect HTPC chip:

• - Super-low idle power consumption. With a little attention paid to the components, it's not hard to build a system that uses less than 20 Watts of power when idle. Low power means low heat, and low heat means you don't need a lot of fans, and can use a small, stylish case (that looks more like AV equipment rather than a PC)
• - Hardware decoding of common video formats. So I can watch my Blu-Ray rips without using a lot of CPU power, thus keeping energy usage and heat to a minimum.
• - Still a "real" CPU for when you need to fallback on software decoding. The nVidia ION based systems share the above two criteria; but with a wimpy Atom CPU, if you ever need a real generic computation ability you're out of luck. The SNB chips give you the best of both worlds.

And I watched in envy as, just after the SNB launch people on forums (AVSForum for example) built Windows-based media PCs that were delightfully capable, small, cool, quiet, and used very little power. When was the SNB launch, like two years ago? And I still can't get vaapi working reliably on my HTPC. My situation is just like this "anyone have vaapi working reliably on sandy bridge" post on Phoronix. Basically, it seems to work for a brief amount of time, but it's not deterministic. Maybe five minutes, maybe a couple days, but ultimately, X will hard lock and the screen will go black. Fixed only by logging in remotely and rebooting. So maybe it works for some. But it certainly doesn't "just work" --- if it does in fact work, there is clearly some magical alchemy of proper versions of all the different software components, and possibly the phase of the moon.

How much shorter is the commute? That alone will add "pay" to the new position. I think a lot of people fail to recognize just how expensive car commuting is. If the new work place is close enough that you can bike, you can really save a lot of money.

These days, when considering what a job pays, you can't just look at the salary. I really recommend reading Your Money Or Your Life . Yeah, the language is kind of mushy and touchy-feely at times, but the general points are important. All for-pay jobs should be considered in terms of their real pay. Which is your salary, minus taxes, minus commuting expenses, minus work incidentals (uniforms or other equipment you personally need to buy), etc etc. Any non-reimbursed expense that you must incur as a result of your job must be subtracted from the advertised salary. IOW, would you spend this money if you didn't have to work? Furthermore, you need to break that real pay into an actual wage, i.e. what is your effective per-hour pay? Take the salary, minus all the expenses I mentioned, and divide by hours spent on work---including your commute, forced breaks, overtime, etc. (So, for example, consider two otherwise identical jobs, but one with different commute times. The one with the longer commute has a lower overall real wage.)

Consider also health insurance benefits. If you're single, it's probably less of an issue. But if you're married and have kids, then it becomes a big deal.

I will say this: I've now had two positions in my career, and in both case I was part of the "expense" structure. In other words, the stuff I worked on was necessary and provided real value to the company, but was not the primary revenue generator. So management views it as an expense, and cost-cutting is the name of the game. How little can we spend and still get the same result? But when you're dealing with a part of the business that is directly responsible for the profits, management tends to be a little more flexible, and willing to take bigger risks. Just something to consider: if you're moving from a position where you work on your company's end product, to one where you are simply part of the "support" structure, you may find the new environment to be frustrating.

FWIW, I was faced with a loosely similar situation: I had a relatively stable job at a big company. It paid a decent wage and I more or less liked it. But from a friend's invitation, I took a chance on a completely new job in a new city at a startup. The startup has been quite successful, and I'm making considerably more money. But I'm not particularly happy with the job itself; not miserable, but it's certainly not something I'd do for free. I stay for the pay. But I don't regret my choice; even if I knew then what I know now, I'd still take the job. The way I look at it, I'm "buying" greater future freedom by sticking with the not-enjoyable-but-high-paying position for now.

My first link in this comment was from Mr Money Mustache, a blog about facilitating early retirement through frugality and saving money. The retirement goal isn't so much of being able to sit around and do nothing, but being financially independent so that you're no longer a wage slave---you can strictly chose what you do based on the fulfillment factor, rather than worrying about putting food on the table. IOW, you can find the job you like so much you'd do it for free.

Do you know anything about the department/group you'd be managing in the new position? What are the people there like? Are they naturally happy and motivated to do good work? Or is it a sweatshop, where your job will be to crack the whip? Are they struggling right now, and just looking for a patsy to take a big fall?

Ultimately it's a personal decision, no matter how many details you provide about each position. I would use Your Money or Your Life (YMYL) to analyze the finances of both positions. In a lot of cases, you'll find that one job has a higher salary only on the surface. But seen through the eyes of someone who's read YMYL, the pay might actually be lower! And after YMYL, do as much work as you can to understand the "feel" and intangible side of the new position. What are the people like? What are management's expecations? What's the culture like?

I think all the hoopla about HFT is way overblown. Rather than try to interfere, why not let it just run its course? It's the hot new industry now, but eventually, that won't be the case. Like most other industries, it's on a trajectory towards commoditization. IOW, increasing effort is required for decreasing profits.

I've worked for a small HFT firm for many years now; helped it grow from startup to where we're at now. Some random factoids that a lot of people don't understand:

• - Everyone always focuses on stocks, but HFT takes place in lots of other markets. My firm barely trades stocks. The big money is in other instruments, like futures (US and foreign), sovereign debt products (e.g. US treasuries), and foreign exchange. And note that while a stock's price is exactly what it is, for most other products, that "price" represents a fraction of the real price. For example, the price of a US Treasury is basically the percentage of $1 million. So even a one-tick movement represents a fairly large change in price.
• - The startup costs aren't that high. Are they trivial? No, but they are not insurmountable. At least my company uses commodity servers, and in some markets, still relatively old networking equipment. The price to co-locate with an exchange is not outrageous. Of course it depends on the exchange, but you can probably get a co-located cabinet for $10k/month or less. Then you add in exchange fees (highly dependent on the venue) and clearing fees. All that stuff is quite unremarkable relative to the start-up costs of any other non-trivial business (dramatically lower than, say building a factory, or buying big earth-moving equipment).
• - I find it amusing that HFT firms claim they benefit everyone by "increasing liquidity". The irony is that they only participate in markets that are already quite liquid. It's hard (virtually impossible) to make money in an illiquid market with low overall volume.
• - The presence of all these automated market-making algorithms ensures a low bid-ask spread. Yes, when people trade, they are effectively paying a "HFT tax". But without HFT, the big-ask spread would probably be much larger, and traders would instead pay the spread. I can't see how people can legitimately complain about this. Remove HFT and you replace the "HFT tax" with the "bid-ask spread tax".
• - Depending on the "style" of HFT, some market-making firms actually, by design, lose money on actual trades. They make their profits from the exchange, who pays rebates for making markets. HFT is a boon for the exchanges, where profits scale with volume.

What kind of people are really hurt by HFT? Certainly not me, as a small-time buy-and-hold investor. Heck, I send market orders. I intend to hold my securities for at least a decade. If I over-pay by a cent or two, I don't really care. And, as I suggested in my last two bullets above, who's to say that I didn't get a better price due to HFT? Maybe some market maker took a loss on the trade with me, but profited overall from exchange rebates? Maybe without automated trading, I would have had to pay a larger bid-ask spread? I don't think anyone can really know the answers to these questions. I'm certainly not losing any sleep over them.

Opposite me are the big institutional investors managing mutual funds, trusts, pensions, etc. These people typically have a goal of meeting some pre-defined position based on the design of their fund. The bigger goal is meeting the fund strategy, rather than micro-optimizing each individual trade. Do they really care about the HFT guys skimming a bit off of each trade? And, as in the case of the small buy-and-hold investor, how can anyone know that they actually are getting a worse price than they would without the HFT guys? How does anyone know that the big institutional investors also wouldn't be subjected to a costly bid-ask spread?

I'm certainly not trying to defend HFT. I keep my job only for the big bonus. I don't really feel like I'm improving society. At the end of the day, I feel like my efforts really just make wealthy people wealthier. But combine the big bonuses with a frugal lifestyle, and I'm on track to retire quite young. That means: more time to spend with my family; and the ability to pick jobs based on how fulfilling they will be, rather than whether or not they pay the bills (I could be a full-time volunteer).

Anyway, I don't think it's worth getting all worked-up over HFT. And as I said at the start of this comment, it will be a commodity industry before long. The other side of every profitable/winning HFT trade is a losing trade for the counter-party. Eventually, HFT methods will come into the hands of these counterparties. Now the previously winning HFT will have to adapt to continue to be profitable. I think at some point, we'll see off-the-shelf trading software that lets people execute their trades with the same sophistication as the HFT shops. In order to remain profitable, HFT will have to keep digging deeper and deeper in their research, and come up with increasingly clever schemes. That's a path for more effort probably less returns. I.e., commoditization. That's the flipside to my first bullet-point above: eventually, it will be very expensive to launch an HFT, and the available opportunities for profit will be shrinking. That makes it overall less lucrative, which implies boring, meaning, you won't care about it any more.

They used to say communism works in theory, but I don't think that old saying still applies. It really doesn't work, if you pick apart the details. I think libertarianism is the same thing: from the big picture perspective, it sounds great, but beyond that, it's logically flawed---it ignores the reality that human nature and human institutions are imperfect and corrupt. Or the idea that the "free market" is the solution to everything---my understanding is that a perfect market is one with a truly level playing field. A truly level playing field is one in which all participants have access to complete information at all times---how is that possible under any circumstances? A perfect market, with a truly level playing field is simply not possible, not even "in theory". So our current system is broken, but the libertarians want to replace it with something that is logically also not perfect.

While I feel that regulations are a double-edged sword, I do think they make my life better. And this goes back to the comment I quoted above: how can anyone think that their life would truly improve if all regulations were removed? Think about how hard your life would be: is the water you're drinking safe? Is your food tainted? Are the buildings in which you live and work structurally sound? Are they wired safely? Are the consumer electronics you're buying filled with mercury or other heavy metals? Are they going to electrocute you when you plug them in? I think the anti-regulation people (libertarians, right?) either don't recognize or at least take for granted how much "free" comfort they get due to regulations. And they always come back and say, but there would be independent businesses, bound by contract law, that would audit foods and structures and consumer products for safety (e.g. Underwriters' Laboratories). But how can anyone think that for profit "regulators" would really work in the interest of the public? They will be paid by the companies making products, and their first priority will always be profits. And how can we audit those private companies, to make sure they are actually honoring their contractual obligations? See my comment above about a level playing field. Furthermore, necessarily, products that are "regulated" by these for-profit companies will cost more. So poor people are further screwed, as they will be less likely to afford the "safe" stuff.

I have yet to have someone explain to me why libertarianism isn't another term for social Darwinism. The fundamental underpinnings of a market economy are supply and demand. What work an individual performs is a function of his personal "supply". That is, his talents, aptitudes, and/or interests. Not everyone can make the big bucks, for one reason or another. As the parent said, most jobs don't pay in proportion to their benefit to society. I just don't get it: how can a pure market economy and a society held together by contract law (isn't that libertarianism in a nutshell?) suddenly change human nature and values in such a way that the needs of the masses are balanced with the needs of the individual? Does anyone really believe that optimizing a society's structure around the needs of the individual will be good for everybody as a whole? Can some people not see (in less that two minutes) that there are plenty of examples where individual interests directly conflict with a group's interests?

At the end of the day, humans are imperfect and therefore human institutions are imperfect. I think the goal of any government is to recognize those flaws, and try to create a structure where the effects of those flaws are minimized. And I don't think there's a single silver bullet to fix it. Like anything we build, a government, a market structure, a society---they all require care and feeding, planning and monitoring, checks and balances. That is, it requires work. I think there's a middle ground where most people will be driven to do that work, for the greater good---that is, work for something that benefits themselves and at worst has a neutral impact on everyone else. And frankly (perhaps in contrast to everything I just wrote), that middle ground is a little less comfort than I think most middle class (and up) people in the USA have come to expect. Too much comfort, and we become shiftless and complacent; too little comfort and we might as well revert back to primitive times where daily life was a struggle to survive. Make life hard enough that I'm motivated to work hard and improve it; but also make life easy enough that I still have the time and energy to care about more than my immediate needs.

I see this mantra repeated often around here, but I'm not so sure it's entirely true. First, what is a "regular stock holder"? On one end, there are small-time, buy-and-hold investors such as myself; on the other end, there are big institutional investors who manage massive portfolios for pension funds, insurance companies, mutual funds, etc. And there's everything in between. From one end of the spectrum to the next, you have very different trading profiles, and thus are affected very differently by high-frequency trading.

For someone like myself, I make maybe a few dozen (relatively) small buys per year. These buys are usually in the neighborhood of 100 shares. If a high-frequency trading program jumps in and effectively front-runs me to to make a few pennies, I don't really care. Overpaying by a penny or two per share means nothing given my buy and hold (long term) strategy. I'm already out $9.99 per trade in commissions to my broker. I'm looking at a horizon of at least ten years, when these relatively small additional costs shouldn't matter.

On the other end of the spectrum is the big institutional investor, like the pension- or mutual-fund manager. This person's job is to constantly rebalance the portfolio to meet some pre-defined metrics; he's generally actively trading huge amounts on a daily basis. While he certainly wants to get the best price possible when he trades, it's practically impossible for him to do that given the volumes in which he deals. Unless he has a highly specialized trading algorithm---that is, something just as sophisticated as the high-frequency traders---he can't help but signal his intentions to the market. Telegraphing his intentions is what makes him a "victim" of the high-frequency traders.

I'm not a fund manager, but my assumption is that, like me and my small buy-and-hold strategy, he also doesn't care about having a small percentage skimmed off of each transaction. To me, it's like buying a big-ticket item, such as a car. Say you budget $27k to buy yourself a new car. Now, some enterprising company goes out and manages a massive, real-time database of every car available for sale in the country. This company can use this database to find you the exact car you want, right now for $27,250. If you're willing to spend $27k, do you really care if you pay an extra $250? And for that $250, you get precisely what you want, and don't have to wait. Compared to going to a dealer, who, if you're lucky, might have what you want at your price... but chances are, the dealer will have something close to what you want, and you'll have to negotiate the price. Or maybe the dealer can get you exactly what you want, but you'll have to wait while he works the intra-dealer process to provision the car. Or maybe he can get you exactly what you want, for even less than $27k, but you'll have to wait for the car to be manufactured. A car buyer can face all these scenarios, but I believe the fund manager most closely mimics the first: that is, he knows exactly what he wants, and he wants it right now.

My prediction is that we'll see the high-frequency trading landscape continue to evolve. Like anything, there will come a day when that kind of business and the skills required to do it are commoditized. And when it reaches that point, it will be much less lucrative. I think we'll see traders of all profiles using ideas and techniques from the high-frequency world in their own trading, meaning that the very people high-frequency traders take from will become direct competitors. The small-time trader like me will implicitly use such techniques, though they will be invisible, as it will actually be implemented by my discount broker (perhaps they'll offer me the BestPrice(tm) service, which just uses high-frequency methods to get me a better price). The big institutional investor will build out his own algorithmic trading system with better "stealth" tactics, and also competes directly with the high-frequency types. I think we are heading towards a truly automated financial landscape. The "market" will consist of a bunch of little competing programs, and we'll see near-100% of all trades being not just electronic, but initiated by an algorithm rather than a human.

The ethics of all this I think is the same as with all technology. In the USA, doctors usually have high salaries. Even mediocre doctors probably fall into the "upper middle class" category, and some exceptional and specialist docs trickle into the "wealthy" category. And I think most folks would agree that an honest doctor's benefit to society is fairly tangible---they're helping you and your loved ones stay healthy. So perhaps they deserve their big paychecks. I also think conventional wisdom holds that jobs like nurse and teacher have tangible benefits to our society, though these are typically "middle class" or even "living wage" positions. But if you can consistently detect a mis-priced asset, or a major market move, and express this as a computer program, you'll find yourself securely grounded in the "very wealthy" class. But where's the tangible benefit to society?

Sometimes I take the socialist viewpoint that this is a major failing of capitalism: wealth isn't distributed according to benefit to society. Of course "benefit to society" is subjective, but our current system says detecting mis-priced asset is more valuable than saving a human life. On the other hand, perhaps that asset mis-price detection is truly more valuable than saving a life, we just don't yet have the ability to demonstrate it.

But ultimately, I think it's just human nature. Whenever there's a technology revolution in any industry, we always see questions of "is it too soon to use this?" For example, vaccines in medicine. As someone expecting his first child any day now, I found that there's a huge body of conflicting information (some of it downright scary) regarding infant vaccination. Just like with high-frequency trading, it seems to be a very polarizing topic, with staunch opinions on either side. I find that in situations like this, the truth is usually somewhere in the middle. And I like to think that my take on high-frequency trading is the middle ground---that it's ultimately neither good nor bad for markets, or even the economy as a whole. There's a lot of money to be made in it while it remains a fad. But as the techniques mature and become commoditized, it will be less glamorous and likewise less controversial. Human history is filled with stories of people making fortunes by being early adopters and implementers of technology. I'm not sure mankind has ever seen a new technology that didn't have a silver lining. Algorithmic finance falls into the same category---great for some, but others see the warts. When high-frequency trading as an industry matures---when finance is overwhelmingly algorithmically driven---it will be the "norm" and the warts will have smoothed over someone or at least accepted.

Who says the infrastructure needs be used efficiently?

Besides, if there was more competition for last-mile connectivity, we'd have 10x the number of providers, meaning, 10x the infrastructure. Sure, the usage of all that combined infrastructure would be "inefficient", but it wouldn't have to be so oversold, because there's the same number of customers spread out over more lines.

As someone else mentioned, fiber, aka "the infrastructure", is cheap. It's getting the rights to bury the fiber that's expensive. And I wonder if that's because so many communities have granted monopoly status to last-mile providers like Comcast?

Another "me too" post: I live in the city of Chicago, and I have exactly two options for consumer-grade wired Internet access: AT&T DSL, and Comcast.

Some parts of the city have more options (e.g., WOW and RCN), but not me.

Why not military boarding schools?

Here's my plan:

1. Install cameras in all classrooms
2. Have monthly "reminder" sessions where the kids learn that they can be moved to a military boarding school for bad behavior
3. After a certain number of reported behavior incidents, the taped footage is sent to some neutral 3rd party (like a jury) to review
4. If the 3rd party decides the reports are valid and the student is disruptive, they are put under military control in a boarding school

So we take out the ones that don't want to be there.

The camera and reporting system could also cut both ways: the teachers should also be held to some standards, and the video can be evidence of under-performing teachers as well. Teachers unions would probably keep this from happening though.

I have family and friends who teach in the public school system, and there are far too many classes where the teacher is nothing more than a baby sitter. Concerned teachers who investigate into the lives of their worst students (either in terms of behavior or academic performance) are usually shocked and appalled at the kids' home lives (or lack there of). My aunt has seen kids whose "bedroom" is like the old school interrogation room: peeling paint on the walls and maybe one light bulb hanging from a cord. Some kids don't even have proper beds. And this isn't the urban ghetto---this is smaller town middle America. How can you expect kids to care about school when their home lives are in such shambles? It sounds like a punishment, but a boarding school would introduce structure and routine into such a child's life.

My mother-in-law has a radical, non-politically correct idea for solving the "lousy parent" problem in public schools: if a child or child's parent is the recipient of any form of public aid, that aid can be withheld for at-school behavioral problems. The more I think about it, the more I like the idea.

Oh, you mean I have to work at it?! Well, then, I'm not interested!

Just joking, I think that makes a lot of sense. I believe I actually witnessed this once: years ago, I went out to lunch with two co-workers. One, call him Fundie, was a fundamentalist Missouri Senate Lutheran; the other, call him Thinky, was a non-denominational Christian. Many of the comments I made in my original post apply to Fundie. Whereas, Thinky is someone I really respect: his faith is very important to him and his family, but he was extremely open-minded, and more than accepting of other viewpoints. To him, church was for being part of a community and providing some structure for his children.

Anyway, Fundie and Thinky started discussing their respective churches. Fundy asked Thinky, "Do you allow gays in your church?" Thinky said that they do allow gays, to which Fundie immediately replied, "But the Bible says it's wrong."

Thinky came back with, "Do you allow divorced people in your church?" Fundie replied that they do allow divorced people, to which Thinky replied, "The Bible also says that's wrong."

Fundie had no response, just sat in awkward silence. I had the tact to restrain myself, but in my mind, I wanted to jump up and hi-five Thinky.

My knee-jerk reaction says you're wrong, wrong! :)

But upon further reflection, though I'm a bit ashamed to admit it, you're right. Though perhaps a bit paradoxical, I suppose it's just as easy to be "blinded by rationality" as it is to be blinded by faith. That is, if you're usually in a rational mindset, you might tend to assume you're always rational. And I think it's inhuman to truly be in a perpetually rational state of mind.

So how do you approach it?

I think you are (at least indirectly) speaking to something that scares the crap out of me: the growing influence of Christian Fundamentalists in the USA. In other words, the people who refuse to believe anything that is incompatible with their faith. By definition, these people are incapable of rational discussion. And when you try to point out their logical errors, they basically say, "that can't be, because the Bible says so," or, as you say, write you off as a fancy pants intellectual. Either way, you are left in a situation where you might as well be speaking two different languages. Actually, if one person refuses to deal in facts and reason, you might as well be speaking to a crazy person, or a dog, or a tree, because the conversation will go nowhere.

I spent the first 27 years of my life in small-town, midwestern USA. I hate to be cliche, but "blinded by faith" quite literally describes a significant number of people I've encountered---within my family, at school, at work, and in the community.

So how do you approach these people, who are either unable or unwilling to communicate rationally? I've thought about this long and hard, but I can't come up with any solution. And I keep seeing suggestions that their numbers, power, and influence are growing. It's conceivable that they will eventually wield some real power (or you could argue they do already). And just as soon as they can, I guarantee you they will try as hard as possible to eradicate all the "fancy pants intellectuals".

I have a friend who teaches 7th grade math at a public school in a small town in central Illinois. She teaches there because, from a student quality and compensation point of view, it's one of the better schools. But the community is small enough that the overwhelming majority of the residents are fundamentalist Christians. Evolution is not taught at this school; school billboards have Christian propaganda all over them; Wednesday is "giving alms" day, and as such, there are no scheduled activities outside of normal classes. On the surface, it looks like a normal public school, but when you get in, you realize it might as well be a private Christian school. And that goes for the community as a whole---on the surface, it is a nice town, mostly upper-middle class residents, low crime, close to a bigger town with all the bigger-town attractions, etc. I always thought it would be a nice place to live until my friend told me about her school. I wonder how many unsuspecting non-Christians end up there, and are quickly run out because of their differences?

In my experience, I've found that what's available varies greatly, even within the same metro area or even city. I used to live in a Northwest Chicago suburb: there were several cable (tv/net) providers, but I lived in a multi-unit building, and Comcast actually owned the building's infrastructure, so they were literally the only option.

I now live in a house in a sparsely-populated part of the City of Chicago, and AT&T and Comcast are the only two options, even though other areas of the city have more options.

I had Comcast cable-based high-speed internet (HSI). I started in on a promo rate that was about $50/month (IIRC) for HSI and basic cable (very basic, effectively local broadcast plus religious, public access and shopping channels). I didn't want the cable, but the promo HSI rate only applied if I got some TV service, so it was actually cheaper to get HSI plus TV rather than HSI alone.

Eventually the promo ran out, and Comcast wouldn't negotiate. I tried calling several times, going straight to the "Cancellations Department", etc etc, and couldn't get them to extend the promo rate. The HSI+TV cost went up to $75/month, and HSI alone was $60/month. To Comcast's credit, the HSI service was generally fast and reliable. But I thought it was too much.

At the same time, AT&T happened to be offering their DSL service for $20/month for the fastest "Elite" tier (6 MB/s). Prices go down from there. I already have a land-line due to having a security system, but I read the promotional material very closely, at it sounded like you could get the same DSL deal without an actual phone line. I was hesitant, but found the DSLReports Forums to be very helpful. One, I was able to ping other forum users to see if there was anyone in my neighborhood with the service (an informal survey of sorts). Two, they have dedicated, private tech support forums where you can actually talk directly to a tech person about the service. In other words, at the cost of waiting a day or two for a response, you actually get a useful answer from someone who knows something, rather than taking your chances with the yahoos at the 1-800 number.

That $20/month is guaranteed for a year, but without any contract (cancel any time), and no setup or equipment fees. I was quite skeptical, but we've had the service for over two months now, and I haven't had any problems. It is slower than Comcast's HSI, but fast enough for streaming HD Netflix, which is our highest requirement.

By the way, my landline is about $16/month, bringing my total monthly outlay to AT&T to $36/month, still cheaper than Comcast. We use cellphones for everything, but need a landline for the security system. If you live in Illinois and have a similar requirement (basic land line service), check out the Citizen's Utility Board. They have negotiated an AT&T plan called "Consumer's Choice Basic", which is as cheap as possible for AT&T. When I signed up for this, it wasn't available online, and I had to call their 1-800 number to get it... and even then, they are shady and try to add services if you don't pay close attention to your order.

As another poster mentioned, there are now really only two chipset vendors for AMD CPUs, nVidia and AMD itself (at least in the desktop space, there may be others in the server space). I haven't done a detailed study recently, but last I checked, both were still sub-par compared to Intel's chipset offerings. nVidia had fairly widespread problems with the actual manufacturing of their chips; it tended to affect laptops more than desktops, but both were at risk. In general, the nVidia chipsets at least used to be quite power hungry.

The AMD chipsets just don't compete with the Intel equivalent. There was a website that benchmarked AMD's SATA 3 against Intel's SATA 2; the Intel won! Also, I don't know if they still exist, but AMD's SATA performance used to change depending on whether you used the native AMD interface or switched to AHCI mode. I forget the details, but you can google for it. AMD also doesn't produce their own ethernet chips. That's not a problem, but it means that AMD boards usually come equipped with an el-cheapo Atheros or Realtek ethernet implementation. I think that Intel makes the best ethernet hardware, hands-down. Yet, outside of the server space, it's impossible to find an AMD board with an Intel NIC on it.

Finally, while AMD's on-board graphics are some of the best you can get (definitely better than anything Intel currently makes), they don't support dual digital output. I have two monitors, and prefer to run them both digitally (i.e. HDMI or DVI). All of AMD's on-board GPUs allow only one digital device to be connected at a time. You can do two monitors with one digital and one analog, but I can see a quality difference; analog sucks.

Overall, though, I think AMD still offers a lot of value. The issues I pointed out will go unnoticed by the casual user. And the power user should understand his needs well enough to know if they make a difference. E.g., I know the Realtek NIC is inferior to the Intel, but is it going to make a difference in day-to-day usage? For the average desktop? Probably not. Another nice feature AMD provides is ECC memory support for all non-Sempron CPUs. With Intel, you have to shell out the big bugs for a Xeon-branded CPU to get ECC support. The unfortunate thing, though, is that most consumer motherboards don't actually provide the traces that allow you to actually use the ECC support! The Biostar A760-G M2+ is a noteable example that actually does let you use the ECC features of the CPU (though it's unofficially supported).

I don't disagree with the overall theme of your post, but just a bit of a nit: I often see people saying I run all these things on my computer, so I need as many cores as possible. The number of processes (or threads) really doesn't mean anything; it's how much work each of those processes are doing. My firewall/router computer is an OpenBSD system running on an AMD Geode processor (500 MHz I think). This is a single core CPU that's slower than even an Atom (I've seen people compare it to a 486!). On this machine, I run a name server, ssh server, time server, DHCP server, and a firewall. The Geode is plenty powerful for all these things for a small home network. Of course, if I was, say an ISP with 100s of customers, the Geode wouldn't be sufficient.

So, in general, any one of the processes you listed can be virtually no CPU load, or require a whole cluster by itself. Consider a mail server. If you're not doing virus scanning, I would guess that you can support 100s of typical users on a pretty wimpy computer, as it's really mostly I/O load. But to run, e.g. gmail.com, that requires a bit more power. The same goes for a web server: if you're serving mostly static pages or relatively simple dynamic pages with a limited number of users (for example, a hobby web development server), it's mostly I/O load, and doesn't take a powerful CPU at all. But, on the other hand, pick any high-profile site, and it's obvious you need a lot more horsepower. The point is, simply saying "I run a webserver" doesn't really actually say anything about the load you're putting on your system.

HTPC = every piece of media you have can be played from a single device.

And DVR capabilities as well, with something like MythTV.

...well, almost. I have the following requirements: streaming Netflix, DVR capabilities, and playback of large-ish DVD rip collection (x264 encoded, stored in MKV container).

A Linux-based HTPC (i.e. MythTV system) can do all of these, except Netflix. I have a PS3, which works for Netflix, but isn't as convenient (have to get up and put the Netflix disc in the PS3). Yeah, that's not a big deal, but the "HT" in HTPC is for "home theater", and to me, having everything accessible from one remote is part of the true HT experience.

I considered running Windows under a virtual machine on my Linux HTPC. While in theory that ought to work OK, it seemed like a "fragile" solution, as the remote control programming would have to be semi-complex to allow seamless switching between MythTV and the Windows virtual machine.

Given the cost (comparable to a Windows license) and ease-of-use of the Roku player, I opted for that.

On the other hand, perhaps a Windows-based HTPC is the real solution? Is there a MythTV equivalent for Windows? That is, a free and reliable DVR with a strong community behind it? Complete with commercial skipping/flagging?!

If I'm not mistaken, XBMC can be used as a MythTV frontend, and is available for Windows.

Now I'm starting to wonder if I was too quick to click "Submit" on the Roku purchase: existing Linux-based HTPC could be moved to the basement, and become a pure mythbackend. Instead of the Roku, I could buy a tiny, ION system for the front room. Install Windows, XBMC, and have the same functionality, but all in one box?

Has anyone else gone this route?