It is actually different from other utilities - the electric company doesn't cap how much electricity you use...

Citation needed. I say that because I believe they DO limit how much electricity you use. Here's the proof: I'm on a residential rate, E01, to be exact. I can't exceed 5kW load under that rate, nor can I exceed 7600kWh per month consecutively while remaining on that rate.

Any other big names I'm overlooking?

I wouldn't think anything special about Apple being missing given that IBM/Lenovo (depending on how the suit would hit based on time of filing), Sony, Cisco/Linksys are missing. A host of others would be potential targets, too.

In 1988, during the height of water cooling technology, 92% of mainframe shops were using water-cooled technology

It's not like water cooling is new.

So the only innovation that should be allowed is innovation qualified through IEEE or another standards body? This way, we immediately have a race to the bottom on price? How would any company ever make money?

So what? If you bought an SLR, you understood the consequences of the connection between the camera body and the lens - or you should have. Where is the limit? Would you propose that GM, Honda, and Toyota all be forced to use the same air filters and fuel filters? What about speedometers and engines? At what point do we accept competition is about solutions and not about making every component in a solution interchangeable?

People buy things that are "locked-in" because they work. If this weren't true, then almost no corporation would have purchased an IP phone system. While there are base standards (SIP), just about every vendor has proprietary extensions that ensure you can actually perform many valuable functions a modern phone system should be able to perform. Note that this does mean the single-source vendor can charge higher prices for additional equipment (i.e. phones), but any company that installs them presumably is saving money vs. their original phone system. So while it's "locked-in", it's still a savings to the purchaser. Decisions are a series of trade-offs, and others will not make the same trade-offs that you make.

...and if that happens, I suspect you'll find it turns into a standard. Lots of things start out proprietary and migrate to standards in order to assist both the manufacturers and consumers. What are now WiFi, HTML, SIP, and many other protocols followed this path. It only makes sense to standardize if the demand and volumes justify standardization.

Then they have product requirements that will lead them to investigate and purchase a different music player. Apple produces a product for a specific segment of the market, they are not required to serve other segments (e.g. those that do care about the headphones)

It's different because anyone who is willing to do that exploration and failure would be considered an early adopter, and not part of the mass market. You cannot successfully sell cheap consumer electronics to only the early adopters.

This is a prisoner's dilemma for manufacturers of hardware that might be "made for X". If any manufacturer performs the qualification, the non-"made for X" product will likely be more expensive. (Yes, I meant "non-", and this is counter-intuitive to some people)

This is all about unit volumes. While it's interesting to target a high-technology crowd with a product, it is *not* the mass-market. The mass-market will flock to the pre-qualified item (even at what would initially be a slightly higher price), and drive unit volumes on the qualified part to the point where manufacturing efficiencies would actually make the qualified item cheaper in the long run.

"Pursuit of life, liberty and happiness" is irrelevant. You're quoting the Declaration of Independence, which is not binding law.

The relevant context is Amendment 4 of the Constitution: "The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated..."

Contrary to the original post, these are probably processor cores, not processors. The article doesn't clarify which is being talked about.

Consider that IBM's original BlueGene was 1M processor cores in 96 racks (and designed in 1997) and 1.6M cores seems in the range of what I'd expect.

I wouldn't. They explicitly choose algorithms that save memory and reduce communications requirements, even if that wastes CPU time. That is rational based on overall system power usage. It is also one of the design characteristics of IBM's BlueGene system, and I would expect that to hold for this system as well.

Actually, many of them have a good basis in logic, but are used beyond their original purpose.

Part of the reason there are individual companies that separated items like brokerages and commercial banking is historical structure created in the Great Depression, known as the Glass Steagall Act

Other responders to your post have pointed out various specific details, e.g. reason for commercial paper, but let me cover a more general point of view on why so many different products exist: Risk.

The issue, however, is that risk doesn't come in only one form. There are different types of risk:
- Default Risk (if a company goes bankrupt, you don't get back your principal)
- Inflation
- Interest rate risk (if interest rates change, then the value of underlying loans change)
- Tax Rate Risk (different tax rates due to different income or time)
- Counterparty Risk (risk of entering into a contract, but the other party failing to fulfill the contract)
- Secured nature of the debt (recovery in case of default)
- Opportunity cost (cost of not doing an alternative with the money)
- etc.

Looking at various products we can see how they are different. An IRA vs. a Roth IRA actually transfers the Tax Rate Risk onto the government (you pay a known tax rate, and the unknown benefit or penalty due to the future difference is absorbed by the government)

A TIPS (Treasury Inflation Protected Security) vs. a normal Bond issued by the government transfers inflation risk onto the government (presumably the normal bond is accounting for perceived inflation in the offering price, but the TIPS accounts for real inflation, thus allowing one to eliminate the risk of the perception of future inflation being incorrect.

We can see today that today's 4 week Treasury Bill Auction resulted in zero yield (give money to the government for 4 weeks, no interest). This presumably would mean the return one could get in a non-FDIC insured bank account (over the current $250K limit) is entirely bankruptcy risk premium.

Also there are organizations that do market clearing of bonds and stocks that absorb counterparty risk. Part of the problem with credit default swaps was that the holders of those products actually bear the counterparty risk, as they are not regulated like other products. (When combined with a lack of market data on quantity and concentration of the risks around default of bonds, this led to one of the underlying issues in the problems we have right now)

Lastly, there are also "positive" values that are priced into different financial products, such as recovery in case of default. That's why secured loans of [statistically] appreciating assets (e.g. home mortgage) are lower rates than loans on depreciating assets (e.g. automobile) and those are lower than unsecured personal loans. Same reason bonds will maintain value longer than preferred stock.

As a specific example of what is good (and bad), look for a moment at interest rate swaps. They actually serve a valuable purpose, in allowing an investor (or the loaning company) to convert a variable-rate instrument into a fixed-rate one (or vice versa). This is valuable to companies to be able to "lock-in" lower interest rates when rates fall, for example. What is risky is when someone speculates on interest rate swaps without having an underlying asset. This results in significant leverage that can be wiped out very quickly if interest rates (in this case) move unexpectedly in the "wrong" direction.

If you lay out the different forms of risk (to both parties in a transaction) and level of security in a loan, I'd suspect you'd find very few financial instruments that significantly overlap (There will be a few, but in many cases, those can be differentiated by size, e.g. commercial paper vs. personal loans). The problems come in speculating on items to make money.

You *can* integrate memory and CPU on the same silicon die, but the overhead to do it in terms of additional error rates and processing tasks makes it economically inefficient to do so. (It is more economically efficient to build a larger cluster). It's possible (I'd even hedge likely) that we will see 3D packaging technology try to get around some of the latency. (More in my post a few years ago). However, the overall difference in speed between memory and processor is unlikely to change in the near future, so we need to continue to architect around that limitation at a system level, rather than a chip level.

The other thing I'd point out is that your analogy to "balanced" general purpose computing systems can (and should) fail for supercomputers... there is no rational reason to continue scaling in a linear fashion.

Then again, this is seriously old news. Trying to optimize a supercomputer to get anywhere close to 100% CPU utilization is known to be a problem. Others have already pointed out IBM's Blue Gene, and there's a reason it's a good example.

From Marc Snir, et al. at IBM, September, 2001 in a file called BlueGenePublic.pdf, which discussed the design philosophy for the Blue Gene supercomputer:

"Standard microprocessors are optimized for running as fast as possible one instruction stream...
Standard nodes suffer from 'von Neumann' bottleneck: computation speed increases much faster than memory access speed"

"Let's think from scratch.... in order to build general purpose systems that overcome constraints of conventional architectures.
Let's accept that significant improvements in cost/performance can be achieved by building an 'unbalanced' system" (emphasis added)

"CPU is a vanishingly small fraction of total system, in silicon area, or power
It is rational to build systems with a surfeit of compute power, so as to reduce memory requirements and reduce the need to move data around
It is cost effective to have a low CPU utilization"