Today's MacIntouch is reporting that some users are encountering kernel panics and have had to reinstall Panther after installing Xcode 1.1. Probably pulled for very good reasons.
I interviewed for an analyst position there several years ago, and although I very much liked the person who would be my boss, I was *very* put off by the head of research. I forget when this was, but Forrester had accumulated a few famous predictions that went nowhere (maybe the eCommerce one), and I asked the question, "how does Forrester learn from the predictions that don't pan out?" I had visions of analysts revising models, tweaking parameters, running new research, etc., just like a scientist in the lab. You know what the head of research said?
"Well, actually, we're nearly always right."
Bite me. That kind of arrogance will kill you, fast. I don't want to join anyplace that isn't a learning organization. And the tone of an organization is set from the top, so I knew this would *not* be the right place to work. Bummer, too, 'cause I otherwise liked the place and the people.
Good link though; I appreciated learning about SketchUp, as I was not familiar with it.
I'm the one who posted this article this morning, and I'm really bummed that the links quit working well before the article actually made it onto Slashdot. Otherwise, I think you (and everyone else) might have seen what is so different about SmoothTeddy: whereas SketchUp looks great for architectural design, SmoothTeddy is better for arbitrary shapes.
The video showed a user drawing an arbitrary closed 2-D curve, and the application would then intuit something about the 3-D surface being represented! Very interesting: suddenly a simple closed curve becomes a potato-shaped closed surface, complete with "pencil-shading" to remind the user of it's 3-D nature. The video continued to show how easy it is to grab and rotate the newly created surface, how drawing a circle on a the surface of the object and adding a new 2-D curve can be used to extrude that surface into a new shape. Cutting, creating indentations, etc., were all demonstrated very powerfully.
Words don't do it (again, pity there's no video), but what distinguishes Takeo's work from SketchUp (and other such apps that i've seeen, in my limited experience) is it's ability to 1) reasonably intuit a 3-D surface based on a 2-D curve, and 2) the simple interface for manipulating that surface / volume.
Every day, the selection of music on iTunes expands in all kinds of directions. For me, I was elated a few months ago when it seemed every great jazz artist ever recorded was "suddenly" they appeared on iTunes--in volume. Try the iTunes Link Maker and run a search to see if your artists are there. Everybody has to make their own choices, but I personally have faith that Apple's clout (and excellent design) will continue to increase the diversity of musical offerings on iTunes.
Oh, I hear ya'! I know the industry seems to be slowly reinventing what LISP was first to do (or Smalltalk, after that), but my point was simply that Java was the first to make virtual machines a successful architectural construct in the marketplace. Before Java, most folks (including many engineers) didn't know what a "virtual machine" was, even though the concept has been around for some time.
BTW: HyperCard and Visual Basic might also be considered an early form of software based on a virtual machine. However, in the case of those products, the virtual machine was never exposed for use outside of the product. Further, both products were marketed as suitable only for specific classes of applications.
With Java (and now.NET), the dialogue has changed: there is now a strong argument that doing *all* user space programming to run on a virtual machine yields high levels of productivity over native code. The advocates of LISP and other such languages have known this for a long time; now the marketplace is catching up.
You know, I like Java technology. If it weren't for the existence of Java, we'd never have this push towards writing user space software in higher-level languages that run on virtual machines. As a productivity aid to engineers, I think it's one of the best advances of the past decade.
However, although Schwartz demo'd some clever technology, it was not very flattering. First of all, it has a little bit of the "me too" syndrome, considering that Mac OS X already has some nice eye candy that uses the same techniques: fast compositing and scaling, to run videos in an icon; translucent windows; windows that easily shift and scale without losing clarity (Expose). Heck, Microsoft demo'd their "me too" six months ago with early images of Longhorn.
Second, was it really necessary to spend the whole time bashing the "dominant" operating system provider. Believe me, I'm no fan of Microsoft, but this anti-Microsoft schtick of Sun's is becoming tiresome, unflattering, and it's not helping their stock price.
I just wish McNealy would try to compete by being better, not by complaining or firing barbs. Frankly, Sun has not been delivering great software technology for several years, so to come at it this way seems very unprofessional. Bummer, too, 'cause I really want to see Sun (and Java) succeed.
I think we're heading to a world where all kernel-space code (on your favorite kernel) is C/C++, and all user-space code is, as you put it, managed code: VM,.NET, or others (e.g., a good open source Smalltalk VM.
Hate to give Microsoft credit, but if they succeed in delivering the vision of Longhorn they recently articulated, then Microsoft may push the industry into this model.
Okay, so I'm a new Mac user who is currently has a pro-Apple bias after replacing my Sony Vaio Linux laptop with a PowerBook. But: Apple has had voice recognition built into the operating system for a *while* as part of their support for Assistive Devices and disabled users. And, btw, the voice support on Mac OS X is seriously good: out of the box, you can control many of your standard applications, just by turning on the speech recognition feature. Sure, the recognizer is not designed to recognize arbitrary sentences and, indeed, uses a state machine model to recognize compound expressions. But, still, how is MS adding this to one of their OS's a big deal? It's not really that innovative--is it?
Could you elaborate? Notes isn't perfect, but it's more secure than many other products of it's class. Of course, it was built in a different era, so I wouldn't be surprised if it was built with keysizes in mind that are no longer "good."
Lotus Notes has long a had a feature to prevent copying or forwarding of messages. With an installed base upwards of 80 million, Notes is one of the most secure e-mail products on the planet, with notable usage among the government and intelligence organizations. Good cryptographic controls are built into the product, so it's easy for individual users to put these kinds of policies in force for their own messages.
And, sure, they've made *huge* sums of money off their operating systems. First they had the deal with IBM for DOS on every PC shipped, then they rolled out the very successful Windows 3.0 (every Windows release before that was a yawner), but it wasn't until Windows 95 that they are really started to get ahead financially with their operating systems.
I guess I always thought that they did applications very, very well, but operating systems not as well. Hence, if they'd just made their application portfolio available *everywhere* (regardless of OS), they could absorb every desktop on the planet.
But, hey, seeing as I would have chosen a different strategy, that might explain why it's Bill with all the billions and not me, huh?;)
I was following the launch event on an IRC channel (via MacNN), and during the course of that IRC someone asserted that Steve said the way they got iTunes onto Windows was by porting Cocoa wholesale--and called it "Yellow Box."
That's a term Apple has used before; IIRC, in the Copeland days, Apple was offering developers it's "Yellow Box" APIs (an early version of Cocoa, I would guess--NextStep wasn't in the picture, though), which would allow them to write to new APIs but with the current Mac OS (Classic) underneath. It was basically a hosting environment, so that once the real OS was released, programs written to the Yellow Box specification would "just work."
I can't confirm that this comment was actually made by Steve Jobs. If he did say this, and he was being serious, then I wonder if Apple now has a framework to let it deliver software on Windows? I don't know about you, I've always wondered why Microsoft never ported COM & a few other things to Mac, Linux, etc., 'cause that would let them leverage their existing codebase on new platforms. Has Apple put itself in a position to pull that trick on Microsoft? Could we see Safari for Windows soon? Or more "insanely great" software on Windows--and not from Microsoft?
Trojan horse might be apt after all; and delivered so innocently, so out in the open at such a cozy event as a music service launch.
As a long time Linux geek (Debian all the way), I switched to a PowerBook this spring and have never looked back. iTunes is a big part of that experience; having it available for my Windows machines at work is even better. Plus, it means Apple has a huge chance at continuing to be viable in the marketplace. After all, this is about selling iPods--not music.
Anyway, having followed the launch even this afternoon and downloading immediately, I can tell you fidelity of the experience on Windows is good: everything is the same as Mac OS X. The look and feel will be recognizable as similar to QuickTime--the brushed metal look so often reviled among older Mac die-hards. Interestingly, I entered the same account information I use on my Mac at home, but that does not allow me to re-download music already purchased onto this machine at the office; if I want it here again (outside of my home network), I need to buy it again.
The Music Store itself appears inside iTunes; it's just another bookmark, like your playlists, your purchased music, any CD you have in your drive, and any other computers on your local network sharing music through Rendezvous. You can play music off another computer with Rendezvous, but you can't add those songs to one of your own playlists, or download / copy them to your machine.
The experience of using the Music Store inside iTunes is a little like a browsere experience, but on steriods: the interface is more sophisticated, but still based on following links for navigation, backward and forward buttons, a home page, etc. On many pages, lists of highlighted albums appear in scrollable horizontal strips of album cover thumbnails. Definitely more than a browser, more than a website.
If you spend time with iTunes, you discover that more and more music arrives everyday. Things you didn't see when you did a search last week are now there. Over time, it starts to have the same jaw-dropping effect as Napster did in it's heyday: all the music you ever wanted, right there.
That's true. In fact, Daniel Kahneman of Princeton won a Nobel prize based on work discovering just that. Essentially, he demonstrates that, contrary to traditional micro-economic theory, the behavior exhibited by actors in the economic arena is not always rational. There are other apparent motivations and descriptors of their behavior. IIRC, John Nash (aka Russel Crowe in "A Beautiful Mind") also won a Nobel Prize with similar discoveries rooted in game theory, but also had important implications for whether or not rationality was the sole descriptor of the behavior of economic actors.
Interestingly, a search on Google for "John Nash Rational Actor" reveals a number of relevant articles, one of which suggests that Nash overstated his discovery's impact on economics.
iChat is for human-to-human messaging; TIBCO's Rendezvous product is for program-to-program messaging as a behind the scenes multicast transport for all kinds of applications. Think "broadcasting real-time stock prices to trader's workstations."
Speaking as someone who was worked 12 years in IT, and has been very successful, I would suggest that it's irrelevant whether or not you find out how to say "no." That's because IT is in the midst of a sea-change, a change that I suspect will ultimately destroy the traditional IT department. You see, it will never be possible to run IT the way you know it could be run; the business won't cooperate enough to let that happen. If you do learn to say "no," vendors, consultants, and suppliers will be right there to tell your customers "yes, yes" no matter what the request is. IT products and services every year are better and better at replacing real, live humans, even developers. You're just one, committed employee; you can't compete against companies that make money off of providing IT services to businesses.
My suggestion? Get out of corprate IT. Go find a supplier, and work for them. Find a way to do what you do but as a contractor, not a full time employee. Create a great product that will destroy the IT departments you currently work for; you'll make a fortune.
Yes, the EJB design pattern helps with clustering by abstracting out many bits of code that make clustering hard (e.g, direct database access); your code simply trusts the app server to manage those bits for you. However, EJB support in an application server of and by itself does not constitute clustering. That's an extra, layered feature that a good app server supports...and JBoss is definitely one of the good ones.
Definitely making some good points, but can I point out that underneath it all e-mail has always been considered an UNRELIABLE technology, yet it is the most succesful internet application ever to this point in history. People have an interest in interacting with one another, and they'll tolerate the lack of reliability at least for a while as a technology matures. Of course, if the reliability of a technology never improves, I wouldn't argue that people wouldn't drop it like the big fad that it was.
Okay, this is making me cranky, so let me use my karma bonus to reply.
We all know that nirvana is hard to achieve, so why are we wasting time insulting eraserewind when *instead* we could be hypothesizing about *how* to head towards nirvana a little more??
And, no, I'm not fucking new here--you probably are, and pretty much ruining it for the rest of us who used to like coming here for insightful discussions about the possibilities of technology.
Criticisms about e. asking for a free lunch, or forgetting economics 101 are missing the point: can wireless technology evolve to a point where our dependency on land-lines is greatly reduced? And can technology be created that accomodates such a world, where every computer is both a transceiver and a relay for traffic?
I would strongly contend the answer is yes. Why? Several trends contribute to the answer.
The rapidly increasing bandwidth and range of WiFi and its derivatives. In less than 5 years, we have seen WiFi move from a fringe technology to mainstream deployment, with the 2nd generation (802.11g, just ratified as a standard) increasing bandwidth by 5-fold.
The increase in applications that exploit peer-to-peer or networked models. The problem with developing networked, distirbuted applications is that they take a different mindset than the ones used to create a single app with a single purpose for a single user on a single machine. As more and more applications adopt these more sophisticated network modesl (e.g. Napster, Gnutella, Jabber, Groove Networks, JINI, JXTA, etc.), the technology will get better.
The number of people who depend exclusively on their cell phones (a related wireless technology), rather than home phones. Such a cultural change will cut into telco revenue--already has.
The number of people who use cable for broadband, not DSL (especially in urban areas). Same as the above, for telcos.
The recurrence of hotspots and "free community networking" as a meme in techno-cultural discourse. Good ideas that don't die prove they have a germ of truth, and only add momentum over time. The final outcome is rarely what everyone would expect (e.g., free wireless for everyone, everywhere), but any good idea that won't go away has proven itself. And who does that impact? The ISPs. When every node is able to negotiate its own entry into the network--who needs ISPs? That was their original function: negotiate the entry to the network.
There's more that would suggest that ISPs and Telcos of the future will either not exist or be radically different, but I haven't eaten my supper yet, so I'm too tired to articulate more thoroughly. It's easy to see that telcos will consolidate around providing high-capacity long-distance links for businesses--wireless will lag beyond land-lines for a long time on both counts will win. And ISPs? In a pervasively networked world, where many nodes are mobile (and many users may switch among multiple, personal nodes), some things have to remain at fixed, well-known nodes--leaving ISPs to consolidate around various forms of hosting and co-location. It may be that in the future, that's what happens to telcos and ISPs: network providers that offer co-location and hosting services.
Is it excessive? That's exactly what Microsoft did when it moved the industry from DOS to Windows 3.0: DOS became just another application running in it's own VM. Of course, in those early versions, one VM could still actually screw up another VM, but with NT, they got it right and wrapped all of 16 bit Windows in a VM. Both users and developers had the option of launching each 16-bit app in it's own VM (i.e., its own copy of Windows 3.x) or in a single system-wide VM shared with other 16-bit apps.
First disagreement: I do think that within 100 years (if not within 10), object-oriented programming (OOP) will wither, if not die completely. OOP as a paradigm was fantastic at coaching us to modularize our previously monolithic code, but we have begun to reach the limits of it's ability to do so.
For example, right now I'm writing a good ol' COM class in C++ and guess how many files have to change, just to add a method? Let's see, that's the.idl,.h, and.cpp. Had I chosen to add a class, I probably would have had to touch another.h or 2, maybe an.rgs or.rc as well. Java is not much better: if you're writing an EJB that leverages Container Managed Persistence, you bounce between interfaces, classes, and XML to make your changes.
The above examples speak to what I mean by modularity, and how I think OOP did help us: it allowed us to localize changes to either a single class or preferably to a single method, instead of modifying code in locations strewn throughout the program. But we've reached the limits of OOP as a mechanism for localization of change; hence, writing OO now for COM or for J2EE, we've lost the benefits of OO.
What comes next? Don't know (yet;) ). My suspicion is that Aspect-Oriented Programming is a step in the right direction, as it allows us to consolidate code into one aspect code that would otherwise appear in multiple classes. This ability to cross-cut our programs, or right code that is aware of it's own program structure is a powerful idea, I believe: Smalltalk has always had it, Lisp and Prolog mostly did, too. In the C/C++ and now Java world, it seems we always write our programs as if the program itself doesn't exist, only the things the program is about. But the ability of a program to exist as a structure itself, and to allow the programmer to reference that program is a powerfully expressive mechanism. I am particularly fond of Prolog, as it need make no distinction between programs and data. Like AOP, Prolog has the ability to allow you to change the behavior of your progaram by just adding a new clause to it's database. No need to go back to the original source code, and find a suitable place for assertion; just add your new rule, and the run-time will accomodate it.
Finally, the second disagreement: I strongly believe parallelism will be fundamentally woven into how we program, rather than an option we add later. In fact, I believe quite the opposite will be true, that most programs will be inherently parallel (or, more precisely, highly concurrent) by default, with pruning of parallelism and concurrency part of the optimization and fault-correction stage. Why this change? The current crop of popular languages still follow the same model as 50 years ago: programs are a single set of instructions executed by 1 machine in sequence from beginning to end, and the program goes away when the sequence reaches the end. Take a look at a C program, tell me that's not how it works, and tell me that most applications today don't still deal with the same basic paradigm.
However, this model doesn't match the resources available to us now, or 100 years from now. Every OS in use has the ability to support multiple threads and multiple processes simultaneously. Many classes of device no longer remain isolated from other devices, but can instead connect through wired or wireless networks, allowing multiple programs on multiple devices to cooperate. The rise of grid computing suggests that programmers are looking to break out of the 1-program, 1-machine paradigm in which we have been trapped. The current crop of grid computing standards are too domain-specific (I would contend), being focused primarily on batch job scheduling and seamless to large, heterogeneous datasets, but they are clearly on to something.
So where's my money? Something like mobile agents written in Prolog with AOP extensions, capable of tr
I think collectively we are all afraid our jobs in IT (or in software, period) are becoming commodities: cheaply paid programmers outside the US are replacing us, open source software is drying up the revenue streams traditionally associated with software.
But it's not all hopeless. There is a way out, a way to prevent becoming the victim of commoditization. There's one skill that almost by definition will never be a commodity, and strangely enough, I had a friend at Microsoft put the idea in my head. The only way to succeed in software (or services, that tag-along so often accompanying software revenue) is by focusing on innovation.
It's that simple. Think about it for a minute: are you maintaining a bank withdrawal application for a large bank, or are you creating protein folding algorithms to run on a massive grid? Are you building the latest revision of the corporation's brochureware website, or are you designing a web-based logistics tracking system for a freight carrier? Are you working for large body-shop, or did you finally decide to start the consulting business you've always wanted? Pick the job opportunities by their potential for tapping into your capacity to innovate, and you'll never go out of style.
Don't give up. Yes, the run of the mill jobs will inevitably go to the cheapest service provider. But innovation is limitless; that's one of the lessons of the '90s that unfortunately seems to have been lost when the money ran out. And it was the money that ran out--creativity doesn't go anywhere. Innovation: do you got it?
Forgive me, this may take a bit. And I do sincerely wish to commend you for thinking through your proposal. However, I wish to share some things from my experience that you may find helpful if you wish to pursue this idea.
I run the IT development wing of a medium size software and services company. I don't have a formal background in financial or managerial accounting, but I've acquired some knowledge in my years, and you would not believe how much I do while acting as the watchdog that ensures our systems (and processes) will not become potential sources of customer fraud. And that means that all our systems (and business models) are built with an eye towards passing a financial audit without raising concern.
Why am I saying this? Because as engineers we sometimes forget that the technical answer to a solution isn't enough; especially the moment you begin taking money from someone as quid pro quo for a service or product you are offering.
When you start accepting money from customers, you must think carefully about the fact that chances are very good that someone has handled a transaction like this before--and been hauled into court for it--or likely will handle a transaction like this in the future and later get hauled into court. Because of that, there are a great number of laws that exist governing how we do business, and there are the Generally Accepted Accounting Principles (GAAP) that provide guidance to businesses regarding the documentation and processes they should have in place to ensure they can properly state their business performance but also demonstrate that they are neither the victims nor perpetrators of fraud.
Almost there, bear with me. So, a micropayment system for bandwidth usage--because bandwidth is such a fungible resource--must have a mechanism that is precise and defensible in a court of law, and it must pass the muster of the CPAs who will eventually come check your books so that you can stay in business. That means:
1) You have to demonstrate you are charging exactly what you said you would charge, calculated the way you said you would. Are you ready to assert your mechanism will always give the right answer, when speaking to a customer (or accountant) who may understand what Apache is?
2) Unless you state clearly why you are not doing so (i.e., you have different services to offer), each customer must be charged the same price for the same thing. Are you sure your system won't accidentally overcharge one customer and undercharge another? What if you customer's compared notes?
3) What if a customer asks you to justify the bill? How would you do that? And if an accountant, 12 months later, asked you to justify the charge to a customer for a specific week of service, what documentation or digital trail would you provide that would precisely show why the charge was what it was?
I could probably go on, causing great boredom to most of Slashdot and you, I'm sure, but I just wish to point out that micropayments are fine, but the systems required to support them are very complex *because* of the requirements like the ones I've listed above. And thus very expensive, with more twists and turns than you've elucidated on a single page.
Again, my commendation to you for proposing an idea for Slashdot to, well, slash. But I couldn't let another post go up detailing a business or technical idea that is still too ungerminated too yet succeed. Best wishes!
Slashdot Mirror
Today's MacIntouch is reporting that some users are encountering kernel panics and have had to reinstall Panther after installing Xcode 1.1. Probably pulled for very good reasons.
I 100% agree.
I interviewed for an analyst position there several years ago, and although I very much liked the person who would be my boss, I was *very* put off by the head of research. I forget when this was, but Forrester had accumulated a few famous predictions that went nowhere (maybe the eCommerce one), and I asked the question, "how does Forrester learn from the predictions that don't pan out?" I had visions of analysts revising models, tweaking parameters, running new research, etc., just like a scientist in the lab. You know what the head of research said?
"Well, actually, we're nearly always right."
Bite me. That kind of arrogance will kill you, fast. I don't want to join anyplace that isn't a learning organization. And the tone of an organization is set from the top, so I knew this would *not* be the right place to work. Bummer, too, 'cause I otherwise liked the place and the people.
Good link though; I appreciated learning about SketchUp, as I was not familiar with it.
I'm the one who posted this article this morning, and I'm really bummed that the links quit working well before the article actually made it onto Slashdot. Otherwise, I think you (and everyone else) might have seen what is so different about SmoothTeddy: whereas SketchUp looks great for architectural design, SmoothTeddy is better for arbitrary shapes.
The video showed a user drawing an arbitrary closed 2-D curve, and the application would then intuit something about the 3-D surface being represented! Very interesting: suddenly a simple closed curve becomes a potato-shaped closed surface, complete with "pencil-shading" to remind the user of it's 3-D nature. The video continued to show how easy it is to grab and rotate the newly created surface, how drawing a circle on a the surface of the object and adding a new 2-D curve can be used to extrude that surface into a new shape. Cutting, creating indentations, etc., were all demonstrated very powerfully.
Words don't do it (again, pity there's no video), but what distinguishes Takeo's work from SketchUp (and other such apps that i've seeen, in my limited experience) is it's ability to 1) reasonably intuit a 3-D surface based on a 2-D curve, and 2) the simple interface for manipulating that surface / volume.
Every day, the selection of music on iTunes expands in all kinds of directions. For me, I was elated a few months ago when it seemed every great jazz artist ever recorded was "suddenly" they appeared on iTunes--in volume. Try the iTunes Link Maker and run a search to see if your artists are there. Everybody has to make their own choices, but I personally have faith that Apple's clout (and excellent design) will continue to increase the diversity of musical offerings on iTunes.
Oh, I hear ya'! I know the industry seems to be slowly reinventing what LISP was first to do (or Smalltalk, after that), but my point was simply that Java was the first to make virtual machines a successful architectural construct in the marketplace. Before Java, most folks (including many engineers) didn't know what a "virtual machine" was, even though the concept has been around for some time.
.NET), the dialogue has changed: there is now a strong argument that doing *all* user space programming to run on a virtual machine yields high levels of productivity over native code. The advocates of LISP and other such languages have known this for a long time; now the marketplace is catching up.
BTW: HyperCard and Visual Basic might also be considered an early form of software based on a virtual machine. However, in the case of those products, the virtual machine was never exposed for use outside of the product. Further, both products were marketed as suitable only for specific classes of applications.
With Java (and now
You know, I like Java technology. If it weren't for the existence of Java, we'd never have this push towards writing user space software in higher-level languages that run on virtual machines. As a productivity aid to engineers, I think it's one of the best advances of the past decade.
However, although Schwartz demo'd some clever technology, it was not very flattering. First of all, it has a little bit of the "me too" syndrome, considering that Mac OS X already has some nice eye candy that uses the same techniques: fast compositing and scaling, to run videos in an icon; translucent windows; windows that easily shift and scale without losing clarity (Expose). Heck, Microsoft demo'd their "me too" six months ago with early images of Longhorn.
Second, was it really necessary to spend the whole time bashing the "dominant" operating system provider. Believe me, I'm no fan of Microsoft, but this anti-Microsoft schtick of Sun's is becoming tiresome, unflattering, and it's not helping their stock price.
I just wish McNealy would try to compete by being better, not by complaining or firing barbs. Frankly, Sun has not been delivering great software technology for several years, so to come at it this way seems very unprofessional. Bummer, too, 'cause I really want to see Sun (and Java) succeed.
I think we're heading to a world where all kernel-space code (on your favorite kernel) is C/C++, and all user-space code is, as you put it, managed code: VM, .NET, or others (e.g., a good open source Smalltalk VM.
Hate to give Microsoft credit, but if they succeed in delivering the vision of Longhorn they recently articulated, then Microsoft may push the industry into this model.
Okay, so I'm a new Mac user who is currently has a pro-Apple bias after replacing my Sony Vaio Linux laptop with a PowerBook. But: Apple has had voice recognition built into the operating system for a *while* as part of their support for Assistive Devices and disabled users. And, btw, the voice support on Mac OS X is seriously good: out of the box, you can control many of your standard applications, just by turning on the speech recognition feature. Sure, the recognizer is not designed to recognize arbitrary sentences and, indeed, uses a state machine model to recognize compound expressions. But, still, how is MS adding this to one of their OS's a big deal? It's not really that innovative--is it?
Could you elaborate? Notes isn't perfect, but it's more secure than many other products of it's class. Of course, it was built in a different era, so I wouldn't be surprised if it was built with keysizes in mind that are no longer "good."
Lotus Notes has long a had a feature to prevent copying or forwarding of messages. With an installed base upwards of 80 million, Notes is one of the most secure e-mail products on the planet, with notable usage among the government and intelligence organizations. Good cryptographic controls are built into the product, so it's easy for individual users to put these kinds of policies in force for their own messages.
It appears the trojan horse theme was picked up on Ars Technica discussions, too, although not with regard to "Yellow Box."
And, sure, they've made *huge* sums of money off their operating systems. First they had the deal with IBM for DOS on every PC shipped, then they rolled out the very successful Windows 3.0 (every Windows release before that was a yawner), but it wasn't until Windows 95 that they are really started to get ahead financially with their operating systems.
;)
I guess I always thought that they did applications very, very well, but operating systems not as well. Hence, if they'd just made their application portfolio available *everywhere* (regardless of OS), they could absorb every desktop on the planet.
But, hey, seeing as I would have chosen a different strategy, that might explain why it's Bill with all the billions and not me, huh?
I was following the launch event on an IRC channel (via MacNN), and during the course of that IRC someone asserted that Steve said the way they got iTunes onto Windows was by porting Cocoa wholesale--and called it "Yellow Box."
That's a term Apple has used before; IIRC, in the Copeland days, Apple was offering developers it's "Yellow Box" APIs (an early version of Cocoa, I would guess--NextStep wasn't in the picture, though), which would allow them to write to new APIs but with the current Mac OS (Classic) underneath. It was basically a hosting environment, so that once the real OS was released, programs written to the Yellow Box specification would "just work."
I can't confirm that this comment was actually made by Steve Jobs. If he did say this, and he was being serious, then I wonder if Apple now has a framework to let it deliver software on Windows? I don't know about you, I've always wondered why Microsoft never ported COM & a few other things to Mac, Linux, etc., 'cause that would let them leverage their existing codebase on new platforms. Has Apple put itself in a position to pull that trick on Microsoft? Could we see Safari for Windows soon? Or more "insanely great" software on Windows--and not from Microsoft?
Trojan horse might be apt after all; and delivered so innocently, so out in the open at such a cozy event as a music service launch.
As a long time Linux geek (Debian all the way), I switched to a PowerBook this spring and have never looked back. iTunes is a big part of that experience; having it available for my Windows machines at work is even better. Plus, it means Apple has a huge chance at continuing to be viable in the marketplace. After all, this is about selling iPods--not music.
Anyway, having followed the launch even this afternoon and downloading immediately, I can tell you fidelity of the experience on Windows is good: everything is the same as Mac OS X. The look and feel will be recognizable as similar to QuickTime--the brushed metal look so often reviled among older Mac die-hards. Interestingly, I entered the same account information I use on my Mac at home, but that does not allow me to re-download music already purchased onto this machine at the office; if I want it here again (outside of my home network), I need to buy it again.
The Music Store itself appears inside iTunes; it's just another bookmark, like your playlists, your purchased music, any CD you have in your drive, and any other computers on your local network sharing music through Rendezvous. You can play music off another computer with Rendezvous, but you can't add those songs to one of your own playlists, or download / copy them to your machine.
The experience of using the Music Store inside iTunes is a little like a browsere experience, but on steriods: the interface is more sophisticated, but still based on following links for navigation, backward and forward buttons, a home page, etc. On many pages, lists of highlighted albums appear in scrollable horizontal strips of album cover thumbnails. Definitely more than a browser, more than a website.
If you spend time with iTunes, you discover that more and more music arrives everyday. Things you didn't see when you did a search last week are now there. Over time, it starts to have the same jaw-dropping effect as Napster did in it's heyday: all the music you ever wanted, right there.
That's true. In fact, Daniel Kahneman of Princeton won a Nobel prize based on work discovering just that. Essentially, he demonstrates that, contrary to traditional micro-economic theory, the behavior exhibited by actors in the economic arena is not always rational. There are other apparent motivations and descriptors of their behavior. IIRC, John Nash (aka Russel Crowe in "A Beautiful Mind") also won a Nobel Prize with similar discoveries rooted in game theory, but also had important implications for whether or not rationality was the sole descriptor of the behavior of economic actors.
Interestingly, a search on Google for "John Nash Rational Actor" reveals a number of relevant articles, one of which suggests that Nash overstated his discovery's impact on economics.
iChat is for human-to-human messaging; TIBCO's Rendezvous product is for program-to-program messaging as a behind the scenes multicast transport for all kinds of applications. Think "broadcasting real-time stock prices to trader's workstations."
Speaking as someone who was worked 12 years in IT, and has been very successful, I would suggest that it's irrelevant whether or not you find out how to say "no." That's because IT is in the midst of a sea-change, a change that I suspect will ultimately destroy the traditional IT department. You see, it will never be possible to run IT the way you know it could be run; the business won't cooperate enough to let that happen. If you do learn to say "no," vendors, consultants, and suppliers will be right there to tell your customers "yes, yes" no matter what the request is. IT products and services every year are better and better at replacing real, live humans, even developers. You're just one, committed employee; you can't compete against companies that make money off of providing IT services to businesses.
My suggestion? Get out of corprate IT. Go find a supplier, and work for them. Find a way to do what you do but as a contractor, not a full time employee. Create a great product that will destroy the IT departments you currently work for; you'll make a fortune.
Yes, the EJB design pattern helps with clustering by abstracting out many bits of code that make clustering hard (e.g, direct database access); your code simply trusts the app server to manage those bits for you. However, EJB support in an application server of and by itself does not constitute clustering. That's an extra, layered feature that a good app server supports...and JBoss is definitely one of the good ones.
Definitely making some good points, but can I point out that underneath it all e-mail has always been considered an UNRELIABLE technology, yet it is the most succesful internet application ever to this point in history. People have an interest in interacting with one another, and they'll tolerate the lack of reliability at least for a while as a technology matures. Of course, if the reliability of a technology never improves, I wouldn't argue that people wouldn't drop it like the big fad that it was.
We all know that nirvana is hard to achieve, so why are we wasting time insulting eraserewind when *instead* we could be hypothesizing about *how* to head towards nirvana a little more??
And, no, I'm not fucking new here--you probably are, and pretty much ruining it for the rest of us who used to like coming here for insightful discussions about the possibilities of technology.
Criticisms about e. asking for a free lunch, or forgetting economics 101 are missing the point: can wireless technology evolve to a point where our dependency on land-lines is greatly reduced? And can technology be created that accomodates such a world, where every computer is both a transceiver and a relay for traffic?
I would strongly contend the answer is yes. Why? Several trends contribute to the answer.
There's more that would suggest that ISPs and Telcos of the future will either not exist or be radically different, but I haven't eaten my supper yet, so I'm too tired to articulate more thoroughly. It's easy to see that telcos will consolidate around providing high-capacity long-distance links for businesses--wireless will lag beyond land-lines for a long time on both counts will win. And ISPs? In a pervasively networked world, where many nodes are mobile (and many users may switch among multiple, personal nodes), some things have to remain at fixed, well-known nodes--leaving ISPs to consolidate around various forms of hosting and co-location. It may be that in the future, that's what happens to telcos and ISPs: network providers that offer co-location and hosting services.
Is it excessive? That's exactly what Microsoft did when it moved the industry from DOS to Windows 3.0: DOS became just another application running in it's own VM. Of course, in those early versions, one VM could still actually screw up another VM, but with NT, they got it right and wrapped all of 16 bit Windows in a VM. Both users and developers had the option of launching each 16-bit app in it's own VM (i.e., its own copy of Windows 3.x) or in a single system-wide VM shared with other 16-bit apps.
For example, right now I'm writing a good ol' COM class in C++ and guess how many files have to change, just to add a method? Let's see, that's the .idl, .h, and .cpp. Had I chosen to add a class, I probably would have had to touch another .h or 2, maybe an .rgs or .rc as well. Java is not much better: if you're writing an EJB that leverages Container Managed Persistence, you bounce between interfaces, classes, and XML to make your changes.
The above examples speak to what I mean by modularity, and how I think OOP did help us: it allowed us to localize changes to either a single class or preferably to a single method, instead of modifying code in locations strewn throughout the program. But we've reached the limits of OOP as a mechanism for localization of change; hence, writing OO now for COM or for J2EE, we've lost the benefits of OO.
What comes next? Don't know (yet ;) ). My suspicion is that Aspect-Oriented Programming is a step in the right direction, as it allows us to consolidate code into one aspect code that would otherwise appear in multiple classes. This ability to cross-cut our programs, or right code that is aware of it's own program structure is a powerful idea, I believe: Smalltalk has always had it, Lisp and Prolog mostly did, too. In the C/C++ and now Java world, it seems we always write our programs as if the program itself doesn't exist, only the things the program is about. But the ability of a program to exist as a structure itself, and to allow the programmer to reference that program is a powerfully expressive mechanism. I am particularly fond of Prolog, as it need make no distinction between programs and data. Like AOP, Prolog has the ability to allow you to change the behavior of your progaram by just adding a new clause to it's database. No need to go back to the original source code, and find a suitable place for assertion; just add your new rule, and the run-time will accomodate it.
Finally, the second disagreement: I strongly believe parallelism will be fundamentally woven into how we program, rather than an option we add later. In fact, I believe quite the opposite will be true, that most programs will be inherently parallel (or, more precisely, highly concurrent) by default, with pruning of parallelism and concurrency part of the optimization and fault-correction stage. Why this change? The current crop of popular languages still follow the same model as 50 years ago: programs are a single set of instructions executed by 1 machine in sequence from beginning to end, and the program goes away when the sequence reaches the end. Take a look at a C program, tell me that's not how it works, and tell me that most applications today don't still deal with the same basic paradigm.
However, this model doesn't match the resources available to us now, or 100 years from now. Every OS in use has the ability to support multiple threads and multiple processes simultaneously. Many classes of device no longer remain isolated from other devices, but can instead connect through wired or wireless networks, allowing multiple programs on multiple devices to cooperate. The rise of grid computing suggests that programmers are looking to break out of the 1-program, 1-machine paradigm in which we have been trapped. The current crop of grid computing standards are too domain-specific (I would contend), being focused primarily on batch job scheduling and seamless to large, heterogeneous datasets, but they are clearly on to something.
So where's my money? Something like mobile agents written in Prolog with AOP extensions, capable of tr
I think collectively we are all afraid our jobs in IT (or in software, period) are becoming commodities: cheaply paid programmers outside the US are replacing us, open source software is drying up the revenue streams traditionally associated with software.
But it's not all hopeless. There is a way out, a way to prevent becoming the victim of commoditization. There's one skill that almost by definition will never be a commodity, and strangely enough, I had a friend at Microsoft put the idea in my head. The only way to succeed in software (or services, that tag-along so often accompanying software revenue) is by focusing on innovation.
It's that simple. Think about it for a minute: are you maintaining a bank withdrawal application for a large bank, or are you creating protein folding algorithms to run on a massive grid? Are you building the latest revision of the corporation's brochureware website, or are you designing a web-based logistics tracking system for a freight carrier? Are you working for large body-shop, or did you finally decide to start the consulting business you've always wanted? Pick the job opportunities by their potential for tapping into your capacity to innovate, and you'll never go out of style.
Don't give up. Yes, the run of the mill jobs will inevitably go to the cheapest service provider. But innovation is limitless; that's one of the lessons of the '90s that unfortunately seems to have been lost when the money ran out. And it was the money that ran out--creativity doesn't go anywhere. Innovation: do you got it?
Forgive me, this may take a bit. And I do sincerely wish to commend you for thinking through your proposal. However, I wish to share some things from my experience that you may find helpful if you wish to pursue this idea.
I run the IT development wing of a medium size software and services company. I don't have a formal background in financial or managerial accounting, but I've acquired some knowledge in my years, and you would not believe how much I do while acting as the watchdog that ensures our systems (and processes) will not become potential sources of customer fraud. And that means that all our systems (and business models) are built with an eye towards passing a financial audit without raising concern.
Why am I saying this? Because as engineers we sometimes forget that the technical answer to a solution isn't enough; especially the moment you begin taking money from someone as quid pro quo for a service or product you are offering.
When you start accepting money from customers, you must think carefully about the fact that chances are very good that someone has handled a transaction like this before--and been hauled into court for it--or likely will handle a transaction like this in the future and later get hauled into court. Because of that, there are a great number of laws that exist governing how we do business, and there are the Generally Accepted Accounting Principles (GAAP) that provide guidance to businesses regarding the documentation and processes they should have in place to ensure they can properly state their business performance but also demonstrate that they are neither the victims nor perpetrators of fraud.
Almost there, bear with me. So, a micropayment system for bandwidth usage--because bandwidth is such a fungible resource--must have a mechanism that is precise and defensible in a court of law, and it must pass the muster of the CPAs who will eventually come check your books so that you can stay in business. That means:
1) You have to demonstrate you are charging exactly what you said you would charge, calculated the way you said you would. Are you ready to assert your mechanism will always give the right answer, when speaking to a customer (or accountant) who may understand what Apache is?
2) Unless you state clearly why you are not doing so (i.e., you have different services to offer), each customer must be charged the same price for the same thing. Are you sure your system won't accidentally overcharge one customer and undercharge another? What if you customer's compared notes?
3) What if a customer asks you to justify the bill? How would you do that? And if an accountant, 12 months later, asked you to justify the charge to a customer for a specific week of service, what documentation or digital trail would you provide that would precisely show why the charge was what it was?
I could probably go on, causing great boredom to most of Slashdot and you, I'm sure, but I just wish to point out that micropayments are fine, but the systems required to support them are very complex *because* of the requirements like the ones I've listed above. And thus very expensive, with more twists and turns than you've elucidated on a single page.
Again, my commendation to you for proposing an idea for Slashdot to, well, slash. But I couldn't let another post go up detailing a business or technical idea that is still too ungerminated too yet succeed. Best wishes!