This is being sold by Staples? Staples sells office supplies, not entertainment. They expect people to pick up a self-destructing DVD while stocking up on printer paper and Post-It notes?
Until recently, solid state storage devices have been treated as "disks". But they're not disks.
They have orders of magnitude less latency.
For files, this doesn't matter all that much. For databases, it changes everything. Solid state devices need new access mechanisms; I/O based seek/read/write is oriented towards big blocks and long latencies. The optimal access size for solid state storage devices is much smaller, more like the size of a cache line. With smaller, lower latency accesses, you can do more of them, instead of wasting channel bandwidth reading big blocks to get some small index item. It's not RAM, though; these devices usually aren't truly random access.
It starts to make sense to put more lookup-type functions out in the disk, since getting the data into the CPU has become the bottleneck. Search functions in the disk controller went out of fashion decades ago, but it may be time to bring them back. It may make sense to put some of the lower-level database functions in the disk controller, at the level of "query with key, get back record". Cacheing at the disk controller definitely makes sense, and it will be more effective if it's for units smaller than traditional "disk blocks"
This could be the beginning of the end of the UNIX "everything is a stream of bytes" model of data storage. We may see the "database is below the file system" model, which has appeared a few times in mainframes, make it to ordinary servers.
Soon, if not already, biotech will be able to create genetically modified humans. But it will take a century or so to tell if a given mod was an improvement. It's going to be a very slow development cycle.
Having done some work on the fringes of the film business, I don't want much to do with it. Projects in development have no money (it took one outfit three tries to send me a credit card number that didn't bounce) and projects in production have no time (they want new features yesterday). People get into film for ego, not money; there are far easier ways to make a few
million dollars.
You get to go to good parties and meet actress/model/waitress types, though.
While the formula may be hugely complex, if such a formula exists, it's kinda self destroying, because the stock market exists in a way because there is no formula.
That's the only part of the above posting that's true. There have been successful technical analysis systems over the years. The trouble is that once someone finds a working strategy for beating the market and uses it on a large scale, others notice and replicate it, and it becomes the market. There's also a failure mode where structured investment vehicles are constructed in such a way that they have a high probability of a continual small gain coupled with a small probability of a big loss, for a negative expectation overall. (See "Long Term Capital Management".)
So much programmed trading activity is going on that it's most of the market now. That's why the number of transactions has become so high.
Here is the
actual paper as a clean PDF. This is the good version.
The linked Technology Marketing Corporation page mentioned in the parent post has only the beginning of the article. It also has 24/7 Media ads in the middle of the article, Google ads on the right, TMC ads at the top, bottom, and in boxes within the article, buttons for more promoted services at the left, a Flash banner at the top, ads from OAS at the lower right, a Digg button, and an email signup box. Oh, and the page refreshes itself every two minutes to change the ads.
The Planet's video tour of why this wouldn't happen
is up and working. Click on "Take the tour", which has many data center pictures. I like the "100% uptime" part.
It turns out they didn't have all the redundancy they said they had. Their central server management system and the DNS servers for those hosts were all in that data center. So customers couldn't get in and switch the DNS to another location for hours.
They now claim to have the server management system back up.
Many young inventors are shocked to discover that you can't just design a part using CAD-CAM and email the design off to a factory in China to be mass produced.
They supposedly had a "short in a high-volume wire conduit." That leads to questions as to whether they exceeded the NEC limits on how much wire and how much current you can put through a conduit of a given size. Wires dissipate heat, and the basic rule is that conduits must be no more than 40% filled with wire. The rest of the space is needed for air cooling. The NEC rules are conservative, and if followed, overheating should not be a problem.
This data center is in a hot climate, and a data center is often a continuous maximum load on the wiring, so if they do exceed the packing limits for conduit, a wiring failure through overheat is a very real possibility.
Some fire inspector will pull charred wires out of damaged conduit and compare them against the NEC rules. We should know in a few days.
Google has clearly gone over to the dark side. They're not just an innocent victim. Google Checkout is laundering spammer money and taking a cut. Google AdWords is accepting ads from spammers. The dark side generates revenue for Google.
From 1917 to about 1980, fear of communism helped keep capitalism in line. During that period, capitalism had ideological competition, and there was a very real fear on the part of business owners that their companies might be nationalized. During that period, most telcos were state-owned. Britain nationalized the steel and coal industries during the 1950s, and most of the rental housing units in the country were state-owned. During the Great Depression, the U.S. Government ran many programs that employed people and built things, a form of socialism.
For over a century, communism was taken seriously as an alternative to capitalism. (Yes, it never worked all that well in Russia. Neither did monarchy, democracy, capitalism, or oligarchy. Russia did better in its communist period than before or since.) During that period, when it faced competition, capitalism had to deliver an ever-higher standard of living.
Which it did. There was more talk of "corporate responsibility" during that period than there is now.
Companies used to fear public opinion. That ended during the Reagan administration. (This is why Reagan was such a darling of business.)
The triumph of unbridled capitalism may be temporary. Something similar happened in 1900-1929, when railroad and power companies ruled the world. That ended in the Depression, and for the next fifty years, businesses were strongly regulated and kept in line.
The next step is to stop treating them as "disks". We tolerate the library and OS overhead of getting to a block on a disk drive because access times on disks are so long. But solid state memory devices can be accessed in microseconds. We need a different model for these devices.
Most of the enthusiasm for "software as a service" comes from companies selling the services. The problem they face is that most companies have already purchased the hardware and applications they need, and they don't need to buy them again. This is Microsoft's big problem. Really, once you made it to Windows 2000 and Word 97, office applications worked pretty well. So why buy them again? Most of the additions since then benefited Microsoft more than the user.
The trouble with "cloud computing" (otherwise known as "time-sharing") is that it only makes sense if you have a transient need for large amounts of compute power. That transient need also has to be at a different time than the transient needs of others. Don't put your retail system on Amazon's system; they have vast excess capacity most of the year, but in November and December, they're busy. (Amazon's primitives for "cloud computing" are well-chosen, though, and they've made some real progress on how to organize large numbers of machines. Their software would be useful without their service, and comparable open-source tools would be valuable.)
If you have a fixed load, but just don't want to run a physical data center, there are many co-location facilities like Rackspace.
How is "cloud computing" different from "grid computing", anyway?
This is lame. They're just putting a plot under an 80s' jumping game.
They should have done one where you're a Saudi prince. You can just play around as a playboy, drive fast cars, and go to the camel races. Maybe buy your own little island off Dubai. Or you can try to work your way up politically. Run a ministry, make alliances with the other princes, try to keep the mullahs happy, the oil flowing, the reformers and the religious extremists under control. Work your way up and try to become King. Maybe try a coup.
Blackberry privacy is only for large enterprises. If you have a corporate Blackberry server, the keys are between the client units and the server, and RIM doesn't have them. If you use Blackberry's public servers, RIM has your E-mail. India only wants "non-corporate emails".
Back in 2000-2001,
our Downside site ran an automatic predictor for dot-com failure. It was amazingly simple and painfully accurate. The system read through SEC filings, extracted the numbers for cash on hand and rate of losses, and projected when the cash would run out. We called that the "death date". That was a good predictor of when the company would go bust. This is a surprisingly good predictor for companies financed via an IPO. You can only IPO once (yes, secondary offerings are possible, but not when you're failing), so there's a finite amount of cash, and when it's gone, so is the company.
For Deathwatch purposes, "dead" was defined as "investors lost essentially all (90% or worse) of their investment". Some of the companies, like Dr. Koop, hung on for years, but their investors did not. (This, by the way, is a common phenomenon to venture capitalists. Many failing companies hang on as overfinanced small companies, downsized until they are able to make just enough money to cover current operating costs but not to recover their startup costs. VC's call these "zombies".) By our standards, essentially all the companies on the Industry Standard list died.
Lee Smolin, the physicist, faced this question a few years back. He'd been encouraging his department to put physics students through programming courses, but became convinced it was a waste of their time. In practice, they were going to use some prewritten package, or have custom code written for them - they weren't going to write it themselves. So he says that "smart people should not program". It's grunt work now.
I mean, when the San Jose Mercury News laid off half of their editorial staff, it really hurt the paper, which went from a top-tier newspaper to almost an advertising throwaway. The Washington Post editorial staff was recently cut to 80% of its peak size, including layoffs of three Pulitzer Prize winners. When the Wall Street Journal was taken over by Murdoch, and key editors were replaced, that was a loss.
These are the people who dig into what the Government and business are really doing. We need them to keep the country honest.
When Wired laid off most of their reporters to become a clone of the Sharper Image catalog, we lost something. The demise of Next Generation magazine, the only gamer magazine with real editorial content, was something of a loss.
But editorial for a second-tier (if that) gamer web site? Who cares? Worst case, somebody spends a few bucks on a crap game and has a boring evening.
It's not a very good article. For one thing, measuring language use by search engine hits is silly. Just because it's the easiest way to get a number doesn't mean anything.
The current situation in programming languages isn't very good. It's been 52 years since Backus invented FORTRAN, and we still don't have it right. Let's look at what we've got.
C and C++ These two languages dominate the hard-compiled field. Yet they're not very good. Every time a computer crashes, the lack of safety in C or C++ is usually at fault. A whole generation of programmers thinks it has to be that way,
although it doesn't; Modula III, Delphi, and Ada were all hard-compiled languages with safety. But their supporting organizations (DEC, Borland, and the DoD Ada center) tanked. C++, incidentally, is unique in having hiding ("abstraction") without memory safety. Nobody tried that before C++, and nobody has repeated that mistake.
C# As a language, C# isn't bad. Unfortunately, it's tied closely to Microsoft's bloated ".NET" environment. The language itself is fast, hard-compilable, and safe. It needs garbage collection, but so be it.
Java An OK language buried under a mountain of mediocre libraries. The language itself isn't bad, but the amount of cruft that comes with a Java project is incredibly large. This may be because the language is free but the cruft is profitable.
Perl It's an awful language syntactically, and it's slow, with a naive interpreter. But it does work well and it's very portable.
Python Python is a nice little language held back by shoddy library maintenance. Python's support libraries (database access, SSL access, web services, etc.) are each maintained by one or two people, with no release coordination. As a result, there's no such thing as a useful "Python distro" that just works out of the box. So web hosting services don't support Python well. Perl handles this little management problem much better, and that's why Python hasn't displaced Perl.
Matlab Matlab? Yes, Matlab. Most engineering computations today are done in Matlab. It's an awful language, and the stock implementation is inefficient, but it's easy to do number-crunching. Matlab is what replaced FORTRAN in engineering.
There's some good work on hard-compiling Matlab for parallel hardware, so the speed problem is being addressed.
Javascript The language of "Web 2.0". How did we let that happen? Amusingly, enough work is going into speeding up Javascript that's it's probably going to be the fastest of the "scripting" languages in a year or two. It's a mediocre language (the Self object model was probably a mistake, and the scoping rules are weird) but it's good enough.
So that's where we are. No really good hard-compiled language in wide use, and the major scripting languages each have some tragic flaw.
All this "agreement" is about is to have a repository for everybody's "ontology" data. It's like SourceForge, only less useful.
Most of what goes in is tuples of the form (relation item1 item2); stuff like this:
(above ceiling floor)
(above roof ceiling)
(above floor foundation)
(in door wall)
(in window wall) ...
The idea is supposed to be that if you put in enough such "facts", intelligence will somehow emerge.
The Cyc crowd has been doing that for twenty years, and it hasn't led to much.
The classic paper on why this idea is bogus is "Artificial Intelligence Meets Natural Stupidity", by Drew McDermott. That was written in 1974, and it's still relevant. There are plenty of citations of this paper on the Web; if anyone can find the full text, please provide a link.
If something like this ever works, it will probably look more like Bayesian statistics than deductive logic.
This was a dumb feature in Javascript. In LISP, there's the "reader", which takes in a string and generates an S-expression, and there's "eval", which runs an S-expression through the interpreter. The "reader" is safe to run on hostile data, but "eval" is not.
In Javascript, "eval" takes in a string and runs it as code. Not safe on hostile data.
JSON is a huge security hole if read with "eval". Better libraries try to wrap "eval" with protective code that looks for "bad stuff" in the input. Some such libraries actually work. Maybe. The process of checking "JSON" input for "bad stuff" is complicated enough that just parsing the input without "eval" can be simpler.
Slashdot Mirror
This is being sold by Staples? Staples sells office supplies, not entertainment. They expect people to pick up a self-destructing DVD while stocking up on printer paper and Post-It notes?
Until recently, solid state storage devices have been treated as "disks". But they're not disks. They have orders of magnitude less latency.
For files, this doesn't matter all that much. For databases, it changes everything. Solid state devices need new access mechanisms; I/O based seek/read/write is oriented towards big blocks and long latencies. The optimal access size for solid state storage devices is much smaller, more like the size of a cache line. With smaller, lower latency accesses, you can do more of them, instead of wasting channel bandwidth reading big blocks to get some small index item. It's not RAM, though; these devices usually aren't truly random access.
It starts to make sense to put more lookup-type functions out in the disk, since getting the data into the CPU has become the bottleneck. Search functions in the disk controller went out of fashion decades ago, but it may be time to bring them back. It may make sense to put some of the lower-level database functions in the disk controller, at the level of "query with key, get back record". Cacheing at the disk controller definitely makes sense, and it will be more effective if it's for units smaller than traditional "disk blocks"
This could be the beginning of the end of the UNIX "everything is a stream of bytes" model of data storage. We may see the "database is below the file system" model, which has appeared a few times in mainframes, make it to ordinary servers.
My horse veterinarian used to use one of those. She had a PC running UNIX back at her office, and used the portable to connect to it from her truck.
Soon, if not already, biotech will be able to create genetically modified humans. But it will take a century or so to tell if a given mod was an improvement. It's going to be a very slow development cycle.
With the 5GB cap, Windows Update alone could use up the quota if you own enough Microsoft software.
Having done some work on the fringes of the film business, I don't want much to do with it. Projects in development have no money (it took one outfit three tries to send me a credit card number that didn't bounce) and projects in production have no time (they want new features yesterday). People get into film for ego, not money; there are far easier ways to make a few million dollars.
You get to go to good parties and meet actress/model/waitress types, though.
While the formula may be hugely complex, if such a formula exists, it's kinda self destroying, because the stock market exists in a way because there is no formula.
That's the only part of the above posting that's true. There have been successful technical analysis systems over the years. The trouble is that once someone finds a working strategy for beating the market and uses it on a large scale, others notice and replicate it, and it becomes the market. There's also a failure mode where structured investment vehicles are constructed in such a way that they have a high probability of a continual small gain coupled with a small probability of a big loss, for a negative expectation overall. (See "Long Term Capital Management".)
So much programmed trading activity is going on that it's most of the market now. That's why the number of transactions has become so high.
Here is the actual paper as a clean PDF. This is the good version.
The linked Technology Marketing Corporation page mentioned in the parent post has only the beginning of the article. It also has 24/7 Media ads in the middle of the article, Google ads on the right, TMC ads at the top, bottom, and in boxes within the article, buttons for more promoted services at the left, a Flash banner at the top, ads from OAS at the lower right, a Digg button, and an email signup box. Oh, and the page refreshes itself every two minutes to change the ads.
The Planet's video tour of why this wouldn't happen is up and working. Click on "Take the tour", which has many data center pictures. I like the "100% uptime" part.
It turns out they didn't have all the redundancy they said they had. Their central server management system and the DNS servers for those hosts were all in that data center. So customers couldn't get in and switch the DNS to another location for hours.
They now claim to have the server management system back up.
Many young inventors are shocked to discover that you can't just design a part using CAD-CAM and email the design off to a factory in China to be mass produced.
Sure you can.
YouTube is back up, after a few minutes of outage, so it was probably a different problem.
YouTube's home page is returning "Service unavailable". Is this related? (Google Video is up.)
They supposedly had a "short in a high-volume wire conduit." That leads to questions as to whether they exceeded the NEC limits on how much wire and how much current you can put through a conduit of a given size. Wires dissipate heat, and the basic rule is that conduits must be no more than 40% filled with wire. The rest of the space is needed for air cooling. The NEC rules are conservative, and if followed, overheating should not be a problem.
This data center is in a hot climate, and a data center is often a continuous maximum load on the wiring, so if they do exceed the packing limits for conduit, a wiring failure through overheat is a very real possibility.
Some fire inspector will pull charred wires out of damaged conduit and compare them against the NEC rules. We should know in a few days.
Google Checkout is the payment scheme of choice for Craigslist spammers. Some other payment processors have kicked off spam tool vendors, but spammers have found a friendly welcome with Google. Google also supplies spammers with free e-mail accounts in bulk. Google's YouTube runs ads and videos from spammers. Google's Blogger provides free hosting for spam and redirection sites. It's full-service evil.
Google has clearly gone over to the dark side. They're not just an innocent victim. Google Checkout is laundering spammer money and taking a cut. Google AdWords is accepting ads from spammers. The dark side generates revenue for Google.
From 1917 to about 1980, fear of communism helped keep capitalism in line. During that period, capitalism had ideological competition, and there was a very real fear on the part of business owners that their companies might be nationalized. During that period, most telcos were state-owned. Britain nationalized the steel and coal industries during the 1950s, and most of the rental housing units in the country were state-owned. During the Great Depression, the U.S. Government ran many programs that employed people and built things, a form of socialism.
For over a century, communism was taken seriously as an alternative to capitalism. (Yes, it never worked all that well in Russia. Neither did monarchy, democracy, capitalism, or oligarchy. Russia did better in its communist period than before or since.) During that period, when it faced competition, capitalism had to deliver an ever-higher standard of living. Which it did. There was more talk of "corporate responsibility" during that period than there is now.
Companies used to fear public opinion. That ended during the Reagan administration. (This is why Reagan was such a darling of business.)
The triumph of unbridled capitalism may be temporary. Something similar happened in 1900-1929, when railroad and power companies ruled the world. That ended in the Depression, and for the next fifty years, businesses were strongly regulated and kept in line.
The next step is to stop treating them as "disks". We tolerate the library and OS overhead of getting to a block on a disk drive because access times on disks are so long. But solid state memory devices can be accessed in microseconds. We need a different model for these devices.
Most of the enthusiasm for "software as a service" comes from companies selling the services. The problem they face is that most companies have already purchased the hardware and applications they need, and they don't need to buy them again. This is Microsoft's big problem. Really, once you made it to Windows 2000 and Word 97, office applications worked pretty well. So why buy them again? Most of the additions since then benefited Microsoft more than the user.
The trouble with "cloud computing" (otherwise known as "time-sharing") is that it only makes sense if you have a transient need for large amounts of compute power. That transient need also has to be at a different time than the transient needs of others. Don't put your retail system on Amazon's system; they have vast excess capacity most of the year, but in November and December, they're busy. (Amazon's primitives for "cloud computing" are well-chosen, though, and they've made some real progress on how to organize large numbers of machines. Their software would be useful without their service, and comparable open-source tools would be valuable.)
If you have a fixed load, but just don't want to run a physical data center, there are many co-location facilities like Rackspace.
How is "cloud computing" different from "grid computing", anyway?
This is lame. They're just putting a plot under an 80s' jumping game.
They should have done one where you're a Saudi prince. You can just play around as a playboy, drive fast cars, and go to the camel races. Maybe buy your own little island off Dubai. Or you can try to work your way up politically. Run a ministry, make alliances with the other princes, try to keep the mullahs happy, the oil flowing, the reformers and the religious extremists under control. Work your way up and try to become King. Maybe try a coup.
Blackberry privacy is only for large enterprises. If you have a corporate Blackberry server, the keys are between the client units and the server, and RIM doesn't have them. If you use Blackberry's public servers, RIM has your E-mail. India only wants "non-corporate emails".
Back in 2000-2001, our Downside site ran an automatic predictor for dot-com failure. It was amazingly simple and painfully accurate. The system read through SEC filings, extracted the numbers for cash on hand and rate of losses, and projected when the cash would run out. We called that the "death date". That was a good predictor of when the company would go bust. This is a surprisingly good predictor for companies financed via an IPO. You can only IPO once (yes, secondary offerings are possible, but not when you're failing), so there's a finite amount of cash, and when it's gone, so is the company.
For Deathwatch purposes, "dead" was defined as "investors lost essentially all (90% or worse) of their investment". Some of the companies, like Dr. Koop, hung on for years, but their investors did not. (This, by the way, is a common phenomenon to venture capitalists. Many failing companies hang on as overfinanced small companies, downsized until they are able to make just enough money to cover current operating costs but not to recover their startup costs. VC's call these "zombies".) By our standards, essentially all the companies on the Industry Standard list died.
Lee Smolin, the physicist, faced this question a few years back. He'd been encouraging his department to put physics students through programming courses, but became convinced it was a waste of their time. In practice, they were going to use some prewritten package, or have custom code written for them - they weren't going to write it themselves. So he says that "smart people should not program". It's grunt work now.
GameTap? Do we, like, care?
I mean, when the San Jose Mercury News laid off half of their editorial staff, it really hurt the paper, which went from a top-tier newspaper to almost an advertising throwaway. The Washington Post editorial staff was recently cut to 80% of its peak size, including layoffs of three Pulitzer Prize winners. When the Wall Street Journal was taken over by Murdoch, and key editors were replaced, that was a loss. These are the people who dig into what the Government and business are really doing. We need them to keep the country honest.
When Wired laid off most of their reporters to become a clone of the Sharper Image catalog, we lost something. The demise of Next Generation magazine, the only gamer magazine with real editorial content, was something of a loss.
But editorial for a second-tier (if that) gamer web site? Who cares? Worst case, somebody spends a few bucks on a crap game and has a boring evening.
It's not a very good article. For one thing, measuring language use by search engine hits is silly. Just because it's the easiest way to get a number doesn't mean anything.
The current situation in programming languages isn't very good. It's been 52 years since Backus invented FORTRAN, and we still don't have it right. Let's look at what we've got.
So that's where we are. No really good hard-compiled language in wide use, and the major scripting languages each have some tragic flaw.
All this "agreement" is about is to have a repository for everybody's "ontology" data. It's like SourceForge, only less useful.
Most of what goes in is tuples of the form (relation item1 item2); stuff like this:
(above ceiling floor)
...
(above roof ceiling)
(above floor foundation)
(in door wall)
(in window wall)
The idea is supposed to be that if you put in enough such "facts", intelligence will somehow emerge. The Cyc crowd has been doing that for twenty years, and it hasn't led to much.
The classic paper on why this idea is bogus is "Artificial Intelligence Meets Natural Stupidity", by Drew McDermott. That was written in 1974, and it's still relevant. There are plenty of citations of this paper on the Web; if anyone can find the full text, please provide a link.
If something like this ever works, it will probably look more like Bayesian statistics than deductive logic.
This was a dumb feature in Javascript. In LISP, there's the "reader", which takes in a string and generates an S-expression, and there's "eval", which runs an S-expression through the interpreter. The "reader" is safe to run on hostile data, but "eval" is not. In Javascript, "eval" takes in a string and runs it as code. Not safe on hostile data.
JSON is a huge security hole if read with "eval". Better libraries try to wrap "eval" with protective code that looks for "bad stuff" in the input. Some such libraries actually work. Maybe. The process of checking "JSON" input for "bad stuff" is complicated enough that just parsing the input without "eval" can be simpler.