Slashdot Mirror
12 Crackpot Ideas That Could Transform Tech
InfoWorldMike passed us a link to an entertaining article with a sort of 'top 12' innovative technologies that could change the world. Some of the techs include solid-state drives, holographic and phase-change storage, artificial intelligence, e-books, desktop web apps, and quantum computing/cryptography. For each of these technologies, expert observers weigh in on the potentials and pitfalls of these disciplines. Here are Esther Lim's comments on e-books: "Another issue, besides the prohibitive cost and cumbersome nature of e-documents, concerns the vast portion of the contracts that were signed and agreed upon before e-books came onto the scene ... That raises questions not just in terms of what rights the user has, but what rights the publisher has vis-à-vis the copyright holder." We've discussed almost all of these technologies on the site at one point or another. Which is the most important? Which one do you think we'll never 'get right'?
A problem with Sony's reader is that when one publisher (Baen) contacted them about software to convert existing ebooks, Sony started talking about wanting royalties per book. So while Baen publishes their books in a variety of formats, don't expect them to publish in Sony's format. But Baen already sees more ebook sales than they do sales to Canada, as an example.
They aren't crackpot ideas, but they are the same tired list of "tech that will change the world" that so-called "experts" trot out every so often so that they can appear relevant and sell more magazine articles and books. My take on a few of these:
e-books have been tried, and they've failed. They will continue to fail until we somehow figure out a way to make an e-book that looks, feels, and behaves exactly like a real book. Good luck with that.
Web-based apps aren't there yet, and will probably never get there until we have protocols that will give you the rich API that coding directly for the desktop will give you. Faking it in Javascript just isn't going to cut it.
True AI has been at least 30 years away for the past 50 years. It's an open question as to whether or not we can ever really get there, or if getting there is even desirable.
The Semantic Web is an interesting idea in theory, but I think the article is hopelessly optimistic thinking that anywhere near a majority of web developers will buy in to it, considering the work involved. It suffers from a chicken and egg problem: it's useless unless everyone buys in, but no one will buy in while its still useless.
As for Project Blackbox, I took a tour of it when it came to the local Sun campus, and it is actually a very impressive piece of engineering. Its uses are probably very limited, but the engineers who worked on it definitely deserve some serious geek points.
Solid state storage could have an immediate impact on computer processing, but its long-term effects are even more important. These include reliability and changes in the way hardware is designed, to take advantage of faster I/O.
With reliability comes complete erasure. Unless your file system or OS incorporates unerase, that could be trouble. A new paradigm for mirroring, such as time-delay mirror algorithms, will be vital
With a really fast fixed storage, bus bottlenecks demand attention. If the speed difference between external and internal storage is less than an order of magnitude, but the I/O bus is too slow to take advantage of that, well, buy stock in motherboard makers with fast busses.
But the big change is to the operating system. All current systems have an implicit distinction between 'RAM' and 'disk': you load a file into memory by opening it. Remove the speed distinction between RAM and disk, and all of a sudden virtual memory schemes lose many of their disadvantages. Faster disk also means dramatically faster database access, so among other things, all of those LAMP-driven blogs will be a lot nicer to troll. Invest in companies selling blog (anti)spam software.
The OS bloat that will result from an all-virtual-memory OS will probably mean eye candy at first, but in the end can take us into a true 3D interface, which will be a paradigm shift as big as the move from text to GUI. Look for a new pointing device, such as a touch-ball or cube (instead of a pad) or a wii-like wireless thingy. Maybe something like riding gloves, that leave the fingertips free to type, or even take the place of both a keyboard and mouse.
sigs, as if you care.
That's a pretty old idea. During WWII, the Allies first studied the practicality of strapping engines on icebergs and using them as aircraft carriers. Finding that unsuitable, a fleet of ships made of ice were actually commisioned. Well, technically the ships would be made of an ice/sawdust mix called picrete(sp?). That would have surely been a sight to see. But the project was scrapped when Canada couldn't build them as fast as they were needed (a near logistical impossibility).
So I don't know that driving icebergs is all that crackpot.
- Queue a file to be copied later to another computer. Why do that, when you can manually make the connection and copy it now? What's the point of adding extra automation that just delays things? Because the extra automation can be used to add routing, addressing, notification, etc. Email.
- View an ordinary formatted text file (maybe a few pictures thrown in), but on another computer. That makes no sense, to rely on an unreliable and slow network connection and on the other computer to be up, when you can just copy the file (or have it emailed to you) so you can look at it whenever you want. Besides, how do you even know how to locate the file? Except the protocol for identifying and exchanging this information allows web applications, and you get the HTTP and the World Wide Web.
- GUIs. Using a little wand or ball or mouse to move shapes around on a screen is okay for specialised applications, but computer data is numbers and words, which are all abstract and have no relationship to things on a screen. Besides, you'd have to give these controller gadgets to everyone in the world with a keyboard already, who wants that expense? Besides, keyboards are always more efficient because you can keep your hands in place. GUIs for real (number and word) applications existed for decades before they caught on.
- Apple iPod - less capable player, relied on PC software for functionality. Well, PC software has a better interface and makes things easier overall - plus the iTunes music store.
I predict the next big thing will be something along these lines. Maybe already here, but dismissed as equally silly.- A display-neutral protocol that lets applications run on a server with the GUI on a user's screen. Not pixel-oriented bandwidth hogs like X windows or remote desktop, but something based on well established GUI components and window layout. Extensible User Interface Protocol (XUP) is a much overlooked example.
- Deductive databases. A reasonable relational database with foreign key constraints means that if you select only the data and tables you want, it should be easy enough for the database to select your joins for you. It's an NP problem, but lots of caching could fix most of that. Oh, plus SQL sucks, and it's nearly criminal that people think SQL and relational database mean the same thing.
- Statistical text analysis. The very beginning has started with SPAM filters and Baysian models. Spammers are starting to figure out how to fight them, but variable length Markov chains have the potential to start to glean more meaning from the text and make better decisions. This could lead to the ability to extract common concepts from phrases or sentences which are different, but mean the same thing. This would allow processing text based on chunks of meaning rather than pattern recognition - far from artificial intelligence, but opens up the possibility of a lot of new very high level applications.
There's a few thoughts. Any other things that seem trivial and with vastly overlooked potential?