Consider also the transition to 64-bit. Intel developed the Itanium. AMD choose to extend i386 with amd64. Intel ended up losing big time - they had to give up on Itanium and start naking AMD-compatible chips. There's no reason to think they won't make a similar mistake again.

Intel was significantly ahead of everyone else. Then AMD provided better performance per dollar even at a larger process size by choosing a better design. Then Intel beat them again. Next, ARM was suddenly outselling both when performance- per-watt became the key yardstick. Things change in the CPU market.

Ten years from now, 64-core processors may be competing against 128-core processors and there's no guarantee that either Intel or ARM would have the best design. Mybe in ten years it'll be all about not RISC vs CISC but EIS, Expanded Instruction Set.

You might not even need a power adapter. The price on this compares favorably to an Arduino, so I might use it where I would have previously used and Arduino. My Arduino projects have never needed a power adapter because they've always borrowed power from whatever they were connected to.

[quote]
You don't need a server. You need a COTS router running OpenWRT and OpenVPN (with hardware acceleration), a couple of well-placed antennas, and a commercial- (not carrier-) grade symmetric DSL, cable, or wireless connection.

In other words: You don't need a million spinning-disks server with its own abilities to serve content, you need a a million low-power NAPs with a gateway to your own content.
How much traffic does google.com see from my small Ohio town of ~45k citizens? Answer: Not enough to swamp a well-proportioned 802.11a link. Or a 45Mbps T3. Or a 75Mbps symmetric DOCSIS connection from TWC...all of which are cheaper than hosting actual servers
[/quote]

You used Google.com as your example. I want to understand what you're suggesting. Are you saying that your router , which is "cheaper than actual servers" is going to serve Google.com search results? It's going hold and query the database of over 5 billion webpages, while doing all of the calculations to rank them for each search term people type in? That's pretty impressive for a little OpenWRT router. If you find a way to do that you'll get really, really rich because right now companies like Google spend hundreds of millions of dollars putting together racks and racks of equipment to be able to rank sort through billions of pages in under a second.

Perhaps that's not what you're saying. Perhaps you're suggesting that you and your neighbors could use wi-Fi or coax to connect to each other, then the neighborhood would be connected to the backbone as usual. I've seen something like that work with television. The neighborhood had one big antenna tower, then there was coax running to each house from the antenna. It was called Community Antenna TV, or catv. Today it's better known as "cable tv".

You see what happens is that in your neighborhood , one family has two Netflix streams running constantly every evening and another guy just wants to check his email. The neighborhood has a 100 Mbps backbone connection, so when a bunch of people try to watch Netflix and Youtube from 6:00 PM - 9:00PM, it gets bogged down. The people just checking their email don't want to pay $80 / month for the neighborhood to have a true gigabit backbone to the internet. Rather, they think the families with multiple Netflix streams should pay their fair share - since they are using ten times as much, they should pay most of the cost. So you end up having different people paying different rates to get different speeds, and someone has to manage all of that. You can hire a company to manage all that for you, making sure everyone is paying their share for the backbone, the shared equipment, line maintenance, etc. The companies who manage all that stuff for your neighborhood are called "ISPs".

It's pretty obvious you've never so much as been in a datacenter, not have any idea how CDNs work (or more _fail_ to work, because few pay any attention to the http spec on proxies).

Some of us actually build this shit and know how it works.

Of the just over 1 billion web sites currently online, fewer than 0.000001% have more than 3 servers per CONTINENT. To have a server in each province / state would increase the costs several thousandfold.

There are about ten web sites in the world that could actually have servers in thousands of locations without going bankrupt.

There is a reason your neighborhood street that you live on isn't 2,000 miles long. It connects to a minor collector (street with several stop signs), which then connects to a major collector (street with a few stop signs), which then connects to an arterial (street with stop lights), which connects to a major arterial (three or more lanes each way), which then connects to a freeway, which then connects to an interstate. Streets are laid out like that because a hierarchy of larger and larger paths is the only halfway efficient way to move stuff from any house in the country to any other house. That's just as true with digital stuff - it only works when you put fat fiber under the rivers, through the deserts, and over the mountains.

Which means someone has to decide where to spend $20 million on the next chunk of backbone, and someone has to fork over $20 million and hope that it's the right technology, in the right place, at the right time, and implemented properly.

I was speaking preflop. Post flop is a whole other ball of wax. Post, you have more information not just about the cards, but from betting. So you can then analyze what the nut is, and who may have it (from their betting), and what it might cost you to showdown.

Yeah that's a restatement of my first-order analysis. At a table of four, you'll have the best hole cards 25% of the time, so you should fold if your cards aren't in the top 25% of possible pairs. There are patterns you can learn to know approximately how many hands beat yours.

Further analysis brings out the fact that sometimes you can call cheap, get a good flop, then have someone bet large into your strong hand. So if you're playing against a table who bets small preflop and large postflop, you might call more often. On the other hand, if there are several potential raisers behind you, calling the one bet might not let you see the flop without calling more, so you should fold more often rather than getting stuck between multiple raises. It just depends on who you're playing against and your position relative to the button.

> I'm sure if I played 20 top level players in heads up, no limit, I would lose against all 20 by a pretty damn large margin.

Perhaps. Actual tournaments aren't generally heads up, they are 8 to a table, with as many tables as needed. In these actual tournaments, you can easily "beat" over half of the players by simply folding anything but the strongest hands. Source - I've actually done this in WSOP events. You can get "in the money" (win prize money) by simply playing super, super tight, folding 90% of the time or more. Winning first place requires switching to a strategy of taking far more risks. Not unlike investing, actually- most millionaires get there by investing in low risk, broad based mutual funds.

  In chess, you CAN beat half the grandmasters while playing heads up against each of them. Simply copy the your first opponent's move against your second and vice versa. They end up playing each other, with you as the messenger.

> The list of skills required is so detailed and complex, it would be very difficult for someone to be a master of everything on that list,

The list of skills isn't things you need to be a master of. In fact, most of the time only about half of the listed words are things you'd be doing on the job. You should, however, know what most of the keywords MEAN. If most of the listed words are in your vocabulary, you can then talk to the hiring manager to find specifically what the job is.

Yeah starting out by calling all the time is STUPID. I think this game has four players, so it'll have losing cards 75%. In the first-order analysis, you should therefore fold 75% of the time. Sometimes you should raise, so you should only call about 10%-15% of the time. With further analysis it gets (much) more complicated, but those percentages are in the ballpark.

Their LEARNING algorithm might be good, it might be very good, but they're starting out with a strategy that sucks. Really sucks. They'd probably be much better off starting with a reasonably good strategy, then learning from there.

I made a simple computer poker bot which worked reasonably well, and should work much better than theirs with learning added to it. I used two phases to come up with a set of starting strategies, then played them against each other to determine the winner. First, I analyzed a few hundred thousand hands of actual online poker, ranking each player's cards for strength. I could then see through simple statistics that one should always raise preflop if your hole cards are XY-s or better, for example. That was stage 1.

In stage 2, I modified that base to create a virtual Phil Helmuth (slow plays, etc) and a few others based on actual champion players, just teeaking the basic statistical strategy to play more like the champ it's emulating. Then play the virtual champions against each other. The two virtual champs who come out on top are the seeds for a genetic algorithm to create the strategy you debut against human players.

I agree this is stupid. Sometimes, though, I like to think of the best arguments I can for the other side's position. In other words, come up with reasons I might be wrong.

In this case, I'd have to admit that ANY time I send an encrypted message, it should always have a way for the good guy to read the message. For example, suppose I use https to send a secure request to bank.com. That must have a way for the good guy, bank.com, to read the message. There's no technical reason it can't be encrypted such that TWO good guys have keys, bank.com and the Good Guy Bureau.

In fact, standard encryption as used by tls does almost that - two people ALREADY have the key which is used to encrypt the message. The sender has the key and so does the receiver. The shared key is then encrypted by another key generated such that two parties can know it, without either ever transmitting it. Mathematically, one could certainly add the GGB key to the algorithm.

  It could be just as unbreakable as the current encryption standards, though those do depend on keys being kept secret. The Good Guy Key probably wouldn't actually be kept secret for long. That's the huge failing that makes it a non-starter from a purely technical perspective- that we'd all be screwed if the FBI's key were ever revealed or cracked. Various attempts at DRM show that widely-used keys are always cracked.

> Supercomputing? HAH!

Of the top 100 supercomputers in the world, 99 run Linux.

I can appreciate what you're saying. I went the opposite way. I use Android, which means I use Google for maps, search, etc. Therefore, I've decided since Google has a good profile of me, I'll try to limit it to ONLY Google, rather than being thoroughly profiled by several different companies.

As a side benefit, Google Now does some pretty cool stuff as their database begins to have good data about my interests and such.

A _design_ flaw that effects all cars of a particular make is likely to result in a class action. A _manufacturing_ defect that effects only your car, or just a few cars, would be an individual suit, not a class action. Also, a class action is often initiated after an individual suit - a plaintiff shows that the defendant owes them, then lawyers put together class action to represent all similarly affected individuals.

Even when there is a class action first, an individually who has been greatly harmed is unlikely to join the class. I just received a check from a class action against Toyota. The check is for $29. Someone severely injured by the flaw wouldn't have joined the class, they sue individually (and maybe settle after filing suit) to get a more appropriate remedy for their specific situation.