The script languages like Perl/PHP/Python/Ruby are dynamic, and fill a role that C++ can never fill. Further, while GC can be added to C++, it changes the programming style so much that it is nearly another language (makes using 3rd party libraries tricky).

Java is a middle ground between C++ and script languages. It has the garbage collection, dynamic class loading, "safe" execution model and extensive libraries like PHP/Python/Ruby, but long running programs like web apps get compiled to optimized native code as they run. Yeah, the language has warts, but it is still more productive vs programmer time than C++.

My daughters have been to Messiah College, PA and George Mason, VA, and NoVa Community College. All three schools have no problem with non-Windows as far as infrastructure (network access), administrative web pages, etc. The big problem has been that teachers will require students to purchase some specific Windows application for statistics or whatever (and not a really pervasive one like Matlab, either - 3 different statistics programs "that real professionals use" so far that I've never heard of). The corresponding courses are short on actually understanding the math, and long on memorizing the menu layout du jour of the app. I wonder if the teachers get kickbacks. The one bright spot was an Arabic course where the teacher used a Java Applet - which works beautifully for all concerned.

is assuming file sharing == illegal file sharing. As others have pointed out, that alone makes the rest of the conclusions meaningless. My stat class said sample size should aim to be around sqrt of population. Of course, smaller samples just lower the confidence intervals. The 1000 people (sqrt(10^6)) is at least in the ball park for a population of 30 million.

The $117K is just the computer hardware. You still need UPS, A/C, Power, and floor space. Add up those, and a reasonable profit, and I'll bet Amazon and EMC don't look so bad. But if you already have the infrastructure, and the marginal cost of adding the storage arrays is low, then the design could save money.

As I understand it, there are hidden variable theories completely in sync with experiment. But they are experimentally indistinguishable from true randomness - and hence serve no scientific purpose (although answering Einstein's famous objection, "God does not play dice"). A hidden variable theory where the "hidden" variables can be deduced by experiment "inside" the universe is no longer a truly "hidden" variable theory.

We've invented a mechanical flea!

Presumably, "opt-in" counts as "in writing", and my library will continue to robocall to announce that my book on hold is available. But on the flip side, I can see all sorts of obscure checkboxes when you order online that enable robocalls should you not notice and check/uncheck them.

Unfiortunately, the advanced civilization living peaceably beneath the surface for millennia, will suddenly begin arming for war.

'Nuff said.

I connected our in house phones to a VOIP adapter from Vonage. Another idea is to use the existing wires to pull Cat-5e through the walls. Start with the farthest away, because the phone wires are likely daisy chained, and you need a full run for each Cat-5e cable. You now have high speed wired network in house. You can even connect selected outlets to VOIP if desired. If you sell, the new owner can use it either way.

The sandbox features are a key feature of J2EE. That is how servlets/webapps/apidujour are isolated from each other. That is how your J2EE app can share a JVM with hundreds of strangers at a hosting facility for lower cost (and lower security) than a Xen/Vmware virtual machine. That is how many people create their own custom multi-app platform to share a JVM that reliably isolates the apps.

Javascript is reasonable in the browser, and has a standard. Flash is an abomination. I let it run only when absolutely necessary (youtube, cough, cough). When I do let it run, it crashes the browser after 20 minutes or so to let me know it's time for bed.

While I agree that they are not as popular, Java applets (for which the sandbox is also key) are much friendlier to the browser, and I trust the Java sandbox a whole lot more than Flash.

While it had been done before, and will probably be done better (or worse), the Java software based sandbox security made a whole class of applications feasible, and enabled a new level of fault isolation for non-malicious software. *nix process based security and its predecessors were ground breaking. Java applies that within a shared address space at the ClassLoader level.

Our proprietary accounting software was built from the get go on open source OSes (starting with IBM EDX - obscure, but open source - though not in the GPL sense). Originally written for a green screen environment, it was abandoned for prettier Windows bases systems by a few clients. All the companies that left functional for pretty either went under or came back. We now have a pretty web front end, and a growing number of EDI, Web Service, and Java interfaces to integrate with other software - even Windows software integrates via SOAP (when there is source available for a Windows programmer to customize).

As far as highly functional open source enterprise accounting software goes, have you looked at Adempiere? I have been playing with it, and it is good enough that my long term goal is to migrate our stuff to its framework.

My favorite simple machines were those used by Native Americans. The spring was one of the most important. From saplings bent to snare a rabbit, to wooden fishhooks that snap open when swallowed, to the wooden bow. I always found it odd that it doesn't appear in the classic list.

"PC filtering software" is not very flexible at all. It ties you to a monopoly OS, and doesn't support filtering gateways (HTTP proxies). Ideally, users should be able to customized the approved list by adding/deleting entries. I would suggest publishing the approved list via secure DNS.