Your logic for ODF is a "lame duck open source clone" is needs more justification.
I wasn't talking about ODF specifically, I was talking about OpenOffice - the file format is pretty much irrelevant. OpenOffice, NeoOffice, StarOffice and all these other free, open-source office clones are still buying in to the paradigm they have inherited from Microsoft - namely, WYSIWYG presentation-based markup. I reject that. As a writer, I do not want to markup "this is italic" or "this is 16pt bold" or whatnot, I want to markup "this is a heading, this is a paragraph etc." For me, whether it's got 1in margins or 2in is irrelevant. The design of OpenOffice is one which puts forward that WYSIWYG paradigm. I don't think in 'pages' or 'character spacing' or 'margins', I think in ideas and how they relate and are organised.
So how do you specify your "XML-based format" if you don't use ODF or OOXML?
Simple. I use an XML-based format that's not ODF or OOXML. Have you seen the size of the OOXML specification? It's like 900 pages long stretched across three different namespaces and Microsoft chose XSD as their schema langauge, even though something like RELAX NG would have been a *lot* saner.
What actual format I use is something I have yet to decide - I've tried DocBook but found it too heavy, and TEI is a little bit too backwards-facing (it's designed for archivists and librarians, it seems). XHTML is kind of light enough for my uses, but there's no good XHTML->FO stylesheets that I've seen, so you are reliant on browsers for turning it in to a PDF.
I may simply specify my own format using a RELAX NG schema and use that. That way, how I write will be determined by me, not by Bill Gates. The XML stack is quite an attractive proposition for me because it's de-facto open - RNG schema, XSL and FO are all viewable, and the document is just text with some extra guff added. Plus it's really, really well internationalized, which is something other data formats struggle with. I can also 'pull-in' other formats like MathML or XForms and embed them within my format.
I feel the same about office software quite generally - it feels inefficient to use it. For instance, I never use spreadsheets. If I have data that needs processing, I'll pop open a terminal window and write a Python script. I have control over the data that is being processed and what happens with it. If I need a graph, I'll type 'import matplotlib' and make one. If the data is not merely temporary, I'll put it up on the web as some kind of data file that others can reuse. That data file will have semantic value of some kind (the elements and attributes will describe where it means rather than just where it is on an Excel grid).
If I need to give a presentation, I'll use Eric Meyer's S5 and a generator script that makes it work with an open source outliner I use. Databases? MySQL (or eXist). The 'office' paradigm does nothing for me, and it's something to be actively worked around. I have access to both Office and OpenOffice on both the Mac and on Linux and I *never* use them. The home-grown ways are just so much more efficient for me.
Now, some idiot is going to say "but you can't expect everyone to just drop Word and Excel and use Python and/or XSL". That may be so. But for me, I'm not going to waste my life working with bad software. If the software I am using sucks, I *will* write my own to replace it. One size does not fit all. With regard to academics, I think that if you are intelligent enough to be splitting atoms or deconstructing 19th century poetry, you are intelligent enough to be able to use a piece of software that's not made by Microsoft. Journals should take the lead here. People actually aren't stupid - almost everyone, however they may fail in school, usually gets a driver's license - because there's an economic motivation to having a driver's lice
XSLT ain't scary! You could always ask over on xsl-list, I'm sure someone may have done some kind of docx->xml->fo->pdf kind of pipe. I did briefly look at the Office Open XML format specification but it was like 900 pages long, and I don't have a decade to write a stylesheet.
The answer is that for all non-trivial purposes, people ought to start using simple, well-specified markup languages and simple transformations. I'm really impressed by TEI, for instance. And if you are a GUI designer, please, please, please find a way of making the XML-based process pretty.
LaTeX isn't just good for equations and other science-specific data - LyX and a decent BibTeX manager (I use BibDesk on OS X) are a great way of keeping large volumes of material well-managed for work in any academic field. Without LaTeX/BibTeX/LyX, I would have probably have never finished my (fairly research heavy) undergraduate dissertation (philosophy).
Now I have switched to XML-based formats and use XSL-FO and Apache FOP to turn it in to PDF/PostScript. I have complete control over the whole process and all of it is reusable, semantic and shareable. Add to that the use of the Web to share data openly, and we could potentially hit a nice sweet spot free of both Microsoft Office and it's lame duck open source clones. Part of the attraction of the open source world is getting away from Word and replacing it with semantic markup where I say what I *mean* rather than say what I want the document to look like. That's why OpenOffice et al. are utterly pointless. Open source should be about replacing bad paradigms rather than just porting.
I can't wait until scholarly journals just sit down and write an XML (RELAX NG?) schema and people use a schema-aware editor to write their stuff.
Cowards? That may be true, but it is irrelevant. I know that in a fist fight, I would probably not win. I'm athletically unfit and not trained to fight (uh huh... that might be why I'm a Slashdot reader). But might does not make right, and any technology that helps make us weedy or fat or unfit guys (and women - hence the thing about rape - men are usually stronger than women, and that's why I'd love to see women be able to defend themselves better) a bit more even in the game of life is good technology, methinks. A gun is just a tool towards that end, and it's one I'd really want beside my bed if I lived in an area where violent attack or burglary was a high risk.
Just as if I have a lot of calculations I would use a computer, if I have a violent attacker coming at me I'd like to have a loaded gun in my hand - it's just a piece of technology that helps solve a problem. The computer helps solve the problem of processing bits of data in series of predefined ways and the gun helps stop humans who are coming towards you with the intent of attacking you. Both can be misused in the wrong hands. But just as governments shouldn't ban the Internet because of child pornography, they shouldn't ban guns on the basis that nasty people will use them but hope that nice people will use them more effectively than the nasty people. A use of an item cannot define that item, however much certain people want it to be so.
Well, perhaps if the cops could actually arrive as a crime is taking place, that might be useful. As might having them on the beat. You still have to pay someone to sit in the little room watching the CCTV cameras - why not pop them out on to the street instead?
Also, if I was living in a south London estate (I don't thankfully, I live down in rural Sussex), I'd rather like to own a gun. The government play this little game where they make 'guns' seem like an evil, evil thing which only south London gangsters would have. I'd like all the respectable, middle class home owners to own guns too. Especially women. Women commit virtually no crime - statistically speaking. If you gave every woman a gun and trained them how to use it, rape would virtually disappear overnight.
How many South London gangstas would be meddling with your house if they knew that the barrel of a shotgun could poke out the letter-box at any moment?
The liberty to not be mugged is tied quite heavily to the liberty to defend yourself using whatever technology you deem appropriate. We do not have that liberty in the UK, and it's a bloody shame that we don't. In the meantime, I'm gonna be keeping a nice, sharp kitchen knife in my bedside table, and you can bet I will defend myself with it. And I'm gonna be looking in to getting a Green Card and going to a state which doesn't see fit to prevent me from defending myself with a gun.
Yes, and that is why the key thing in the Semantic Web space is the making of assertions. If I say that your page is about fishing and you say that your page is about fishing, that's no problem. If you say that your page is about fishing and 90,000 people say it's junk, that's when we have a problem. Having a data model like RDF which allows these assertions to be made across the web is what makes such a thing possible without it all becoming too centralised to be useful.
As the old adage says: if voting did anything, they would have outlawed it already.
Not to poke my nose in to French politics, but it's a perfect example of what I'm saying - both Royal and Sarkozy fit my descriptions of politicians down to a tee. I saw Sarkozy speak last December in Paris and he's a control freak nutbag, a wannabe conservative who'll say all the right-sounding things on crime and immigration but won't actually make our lives less regulated. I imagine Royal isn't much better, although she'll just say all the lefty-sounding things on human rights and welfare but won't actually make us any freer. Both will probably be introducing legislation that'll make any sane Internet user scream and will do nothing to bring technology investment or innovation to Europe. Still, at least YRO will have some fun new stories that it can cover from France if Sarkozy wins.
Elections are once every few years. The polling station is 30 minutes walk or about three minutes drive. It's open from 7am-10pm. The physical process of voting is pretty easy. When I used to live up in the Midlands, it was maybe a ten minute walk to the polling station.
What needs to be fixed is I want someone I can vote for who isn't a self-aggrandizing, liberty-stomping, populist arsebag who will sell out my freedom for whatever trinkets of tabloid approval are currently in vogue. Much as I love the Internet, it is not likely to make our political leaders any less dishonest. They're all fucking useless shits regardless of whether my vote is delivered over dead-tree or TCP/IP.
Plus, I live in one of the most rigidly Conservative-voting constituencies in the country, so my vote is almost totally irrelevant. I'm not sure I'm going to be voting on Thursday.
Yes, it's typical "argh! It's not happening!". Well, unless people take an interest and do something, of course it won't happen.
The adoption problem needn't be. If companies and organisations are unwilling to put data up semantically, someone else will. We see this already with accessibility - in the UK, a replacement train times website has been built to replace the crappy National Rail website. I wrote a MySpace screen scraper recently so I wouldn't have to visit the profiles of my friends, and instead I could subscribe to RSS feeds. Semantic data will be available either officially or unofficially quicker than you would think...
So... you let people make their own ontologies. Since the Semantic Web is an open platform with open standards, this is easier than you would think. The only objection to this is "people are stupid". I don't think they are. YMMV.
The Semantic Web layer cake isn't a technology stack per se. There are many ways of implementing each of the pieces in the layer cake.
The following things are what I'd say are currently necessary: URI/IRI, XML and namespaces, the RDF model. Query has been implemented quite neatly with SPARQL but it doesn't require that one have ontologies already in place. Yes, ontology-aware SPARQLing is useful as it solves the Tower of Babel problem to some extent, but it's not necessary.
How we get to Trust does not have to be through Signature and Encryption. For instance, OpenID as a part of the Semantic Web is something I'm very interested in, and I think will play an important part - certainly in the application I'm building.
The semantic web is: a webby way to link data. That is all. Everything beyond that is entirely optional fluff: data vs metadata, syntaxes, ontologies, query languages, rules, logic,...
Also, a lot of the parts are fairly immune to breaking. URIs don't 'break' that easily. I can't quite imagine what the alternative would be. The whole thing written in C? Maybe Assembly? Doesn't solve the design goal...:)
That sounds just like a big user experience problem. There is nothing technically wrong with the data model, it just needs to have people design decent, usable interfaces and make the experience of using the system easier. That, combined with motivation - putting incentives in front of people to make data available. Sounds enough to keep the Silicon Valley type startup culture rolling for a few more years...:)
The flipside to this is that the more 'varied' user agents are, the less susceptible they are to exploits. The Semantic Web - broadly understood - could find solutions to SEO. Search personalization and recommendation systems can go some way to neutralize SEO. My inbox on del.icio.us gets no spam at all, because the only people who can post things to it are people I trust. A web based on 'subscriptions' and pubic expressions of 'trust' will be a lot harder to spam and SEO.
Have you seen GRDDL? It's a way of producing RDF/XML data from "plain XML" (although not plain CSV...), and XHTML too (and HTML if you run stuff through Tidy). It's typically W3C - a big long document to explain 'XML+XSLT=RDF', but it's pretty neat nonetheless. Anything that's even vaguely structured can be turned in to RDF very easily. The W3C made a mistake in thinking that RDF etc. would be picked up straight out of the gate in 1999 (when RDF was standardised) - instead, it's taking a rather bending path towards popularity.
There is lots of interesting stuff going on just below the surface, but because it is going on below the surface, it's very easy for people to say there's nothing going on. I have to strongly disagree with Stephen's post. If the MySpaces of the world don't want to expose APIs, that's their prerogative. It'll be uppity programmers who'll write screen scrapers to get them on the semantic web whether they want to be there or not! The market will ebb and flow, and eventually MySpace will be as dead as all the other social networks that sit on the scrapheap of the 'net. Software competes on a lot of things - features being one of them. On the Internet, a site which lets you export your data will be more valuable to the (intelligent) user than one that doesn't. As for the stupid user, well, that's a problem that the Semantic Web can't solve. Stupidity, like lying, is a problem that no amount of W3C standards can solve.
Whitelisting and search personalization. When your search engine is giving priority to people you know and trust and sites they say they trust then the spam problem is significantly lessened. The OpenID community are already using OpenID as a spam whitelisting mechanism. In the Semantic Web layer cake, there is a layer called Trust, which is based on a combination of the RDF stack and document signing and encryption. Even if that isn't the way it's approached, Trust is something that SemWeb people are concerned with. I'm in the middle of designing a SemWebby tool (can't explain it here because I haven't decided whether it's going to be free, commercial or open source etc.) and how trust is dealt with is important.
Currently, Google is the thing to spam - you pick some keywords and you try to spam them. A highly personalized version of Google that uses your friends and colleagues attention data will be a lot harder to spam. With the Semantic Web, yes, people will lie. But if people find out they lie, they can put them on a "liar's list" - just as you do mentally in real life. People lie in databases, people lie on their blogs, people lie in real life. If you expect the Semantic Web to solve a common trait that is present in a large chunk of humanity, then you need to rethink your assumptions about the Semantic Web.:)
Mod parent up. I spent just under a year in the art school at DMU. What a tremendous waste of time. The technical skills I could have taught myself for a lot less than it cost to there. I have a lot of respect for the old concept of polytechnics and vocational education. But these places try too hard to go in to pure science and humanities, and the institutions that come out of the process are a joke. Similarly, the government are trying to implement "professional studies" modules in to more and more of the curricula. Fortunately, we haven't yet had "professional studies" in the philosophy departments, but I can't see it being too far off. British universities get more and more ridiculous by the day.
How about the people who are at the other end talking to this person trying to persuade them that, yes, I have tried plugging it in and it still doesn't work, and that, yes, you should bump me up to someone who actually knows what they are talking about...?
Oh, what we pay our 49 pennies a minute for...
You'll get no disagreement here
on
PHP 5 in Practice
·
· Score: 2, Interesting
I've gotta agree with some of the presuppositions and points in this review. PHP - in the right hands - is a powerful language. It's great that you can whip things up quickly. But too many of the books go through the fairly simple bits in mind-numbing detail (like, I know what an array is). Thanks to Eclipse and oXygen, most of the time that I'm at a screen, my editor can load the documentation in to a panel while I'm typing it (along with remembering class names, variable names and so on). If I'm coding PHP (or a lot of other languages, for that matter), Eclipse has the language reference. And oXygen gives me the documentation from XSD/RNG/DTD schemas for XML/XHTML etc. Language references aren't useful in dead tree format.
But something that dead tree can be useful for is conveying development experience. Of course, this can be transmitted in other means. Books that give me best practice guidelines are often far more useful than language guides. For instance, in PHP, there is a function called file_get_contents(). It does what it says. You give it a URL or file and it reads it in to a string. But what the language reference *doesn't* tell you is that for getting things off the 'net libcurl is a better way of doing it - it's quicker, more powerful and has a lot of extremely useful options - in short, something which, if one is intent on building a serious web application in PHP, one should probably use. This is one of the reasons why I think language references would be better if managed on a wiki - or, as PHP does it, with comments attached. That way, people can post code samples, bug reports, workarounds, common errors and so on. This is useful.
A measure of a successful technology book these days has to be "does this make me a better developer?". The fact that we have books which deal with best practice means that online documentation has gotten better and better and publishers are responding to that. Most of the languages and frameworks I use I carry the specifications for on my Palm Pilot in Plucker format. Reference books can't compete with that. A few publishers I've seen are shifting towards a tutorial style (in the web design sphere, Friends of Ed is a good example of this).
There are other reasons why web apps tend to be useful:
1. Access from multiple machines (there are some protocols that can handle this reasonably well - IMAP ad Jabber, for instance - but the Web does it without fuss). 2. Open APIs to allow people to reuse existing data in new ways (eg. taking the Chicago crime statistics or the London Tube data and plotting them on a Google Map). 3. Upgrade cycle can be a lot easier - an upgrade can be as easy as "put class.main.php" or "svn commit" etc. Bug-fixing can be done in a similar manner.
That said, having applications that access remote servers is not "just recreating HTML/ Javascript". There's a reason why Amazon has an API to search their products and perform shopping-basket applications - it's so that developers can implement Amazon functionality in to applications - web or otherwise - that could benefit from it. I've been building a web application to allow people to explore each other's music collections. Here, Amazon's data is extremely useful - as I can point people from my application to go to Amazon and buy the CD. There are even classes available in Perl (DBD::Amazon and various others - search for 'Amazon' modules on CPAN), PHP (Services_Amazon via PEAR) and Python (Google for 'pyamazon', I think).
You say that it would be bad for each web site to have to create a desktop application. That's not where things are going. What needs to happen is for each of the companies to release a sensible, sane API, offer some motivation (affiliate programme is one way), and the apps will follow.
Do you think my parents are going to be telnetting in to their mail server? Seriously.
One of the good things about Ajax is that it's asynchronous, which means that it can work on unreliable connections reasonably well. I can open up Gmail on my laptop while I'm at home on my DSL connection. Then I can shut my laptop, open it on the commuter train where my connection is an unreliable mobile phone modem connection and, though the service may come in and out, I don't need to log in again. I use SSH a lot, and it's almost impossible to use over an unreliable modem connection. Most of the better programmed Ajax apps run fine though.
If you want to replicate this experience, grab yourself a GPRS connection (I dunno what they call that over in America, but it's not EVDO, it's the slow one that is used on the Treo 650 etc.) and try logging in to a telnet or SSH connection.
There *are* promising developments to solve this problem. One is Adobe Flex, which will allow developers to release simple web applications for the desktop that sync up with online servers. Another is XForms, an XML specification for forms, which could be combined with, say, a Java application to make very simple GUI applications. The final way is just to wait. Connectivity will be everywhere eventually. Even railway tunnels.
Please come and work at my university. We need sane systems people. We have a Windows + wi-fi setup. It sucks. It's slow. The filters are extremely badly configured - the letters "mp3" are filtered from all URLs (no distinction between 'download a Jay-Z track' and 'download an MP3 of an academic debate'). The deskop environment is locked down so that the only applications that a user can use are Word and Internet Explorer. I wanted to work on an ASCII file the other day and Notepad is filtered. I shit you not.
I'd make a big fuss about it, but I finish in a couple of months and most of the time I can get around the filtering (SSH in to web server, wget/links - might set up a tunnel if I get bored).
How about "so easy, even middle can do it"? Or "so easy, even George Bush could do it".
That said, having spent a few days trying to get Java to install properly on my Ubuntu machine so that I can run oXygenXML, it's *not* easy. Multiple versions not working with one another, different dependencies. It should be pre-installed, or at the very most "sudo apt-get install java". Perhaps I'm better suited to a management position...:)
Yes, it's a masterful move on the part of Microsoft's PR department.
If you want to satisfy large government and corporate buyers who want the data security that XML/(your favourite open data interchange standard) brings, you have three options:
1. Make a sensible, well-designed XML standard to facilitate interop. 2. Don't bother. 3. Make an insanely complex XML standard to facilitate interop in theory while hindering it in practice.
Microsoft have chosen the last option. Anybody surprised?
Slashdot Mirror
I wasn't talking about ODF specifically, I was talking about OpenOffice - the file format is pretty much irrelevant. OpenOffice, NeoOffice, StarOffice and all these other free, open-source office clones are still buying in to the paradigm they have inherited from Microsoft - namely, WYSIWYG presentation-based markup. I reject that. As a writer, I do not want to markup "this is italic" or "this is 16pt bold" or whatnot, I want to markup "this is a heading, this is a paragraph etc." For me, whether it's got 1in margins or 2in is irrelevant. The design of OpenOffice is one which puts forward that WYSIWYG paradigm. I don't think in 'pages' or 'character spacing' or 'margins', I think in ideas and how they relate and are organised.
Simple. I use an XML-based format that's not ODF or OOXML. Have you seen the size of the OOXML specification? It's like 900 pages long stretched across three different namespaces and Microsoft chose XSD as their schema langauge, even though something like RELAX NG would have been a *lot* saner.
What actual format I use is something I have yet to decide - I've tried DocBook but found it too heavy, and TEI is a little bit too backwards-facing (it's designed for archivists and librarians, it seems). XHTML is kind of light enough for my uses, but there's no good XHTML->FO stylesheets that I've seen, so you are reliant on browsers for turning it in to a PDF.
I may simply specify my own format using a RELAX NG schema and use that. That way, how I write will be determined by me, not by Bill Gates. The XML stack is quite an attractive proposition for me because it's de-facto open - RNG schema, XSL and FO are all viewable, and the document is just text with some extra guff added. Plus it's really, really well internationalized, which is something other data formats struggle with. I can also 'pull-in' other formats like MathML or XForms and embed them within my format.
I feel the same about office software quite generally - it feels inefficient to use it. For instance, I never use spreadsheets. If I have data that needs processing, I'll pop open a terminal window and write a Python script. I have control over the data that is being processed and what happens with it. If I need a graph, I'll type 'import matplotlib' and make one. If the data is not merely temporary, I'll put it up on the web as some kind of data file that others can reuse. That data file will have semantic value of some kind (the elements and attributes will describe where it means rather than just where it is on an Excel grid).
If I need to give a presentation, I'll use Eric Meyer's S5 and a generator script that makes it work with an open source outliner I use. Databases? MySQL (or eXist). The 'office' paradigm does nothing for me, and it's something to be actively worked around. I have access to both Office and OpenOffice on both the Mac and on Linux and I *never* use them. The home-grown ways are just so much more efficient for me.
Now, some idiot is going to say "but you can't expect everyone to just drop Word and Excel and use Python and/or XSL". That may be so. But for me, I'm not going to waste my life working with bad software. If the software I am using sucks, I *will* write my own to replace it. One size does not fit all. With regard to academics, I think that if you are intelligent enough to be splitting atoms or deconstructing 19th century poetry, you are intelligent enough to be able to use a piece of software that's not made by Microsoft. Journals should take the lead here. People actually aren't stupid - almost everyone, however they may fail in school, usually gets a driver's license - because there's an economic motivation to having a driver's lice
XSLT ain't scary! You could always ask over on xsl-list, I'm sure someone may have done some kind of docx->xml->fo->pdf kind of pipe. I did briefly look at the Office Open XML format specification but it was like 900 pages long, and I don't have a decade to write a stylesheet.
The answer is that for all non-trivial purposes, people ought to start using simple, well-specified markup languages and simple transformations. I'm really impressed by TEI, for instance. And if you are a GUI designer, please, please, please find a way of making the XML-based process pretty.
LaTeX isn't just good for equations and other science-specific data - LyX and a decent BibTeX manager (I use BibDesk on OS X) are a great way of keeping large volumes of material well-managed for work in any academic field. Without LaTeX/BibTeX/LyX, I would have probably have never finished my (fairly research heavy) undergraduate dissertation (philosophy).
Now I have switched to XML-based formats and use XSL-FO and Apache FOP to turn it in to PDF/PostScript. I have complete control over the whole process and all of it is reusable, semantic and shareable. Add to that the use of the Web to share data openly, and we could potentially hit a nice sweet spot free of both Microsoft Office and it's lame duck open source clones. Part of the attraction of the open source world is getting away from Word and replacing it with semantic markup where I say what I *mean* rather than say what I want the document to look like. That's why OpenOffice et al. are utterly pointless. Open source should be about replacing bad paradigms rather than just porting.
I can't wait until scholarly journals just sit down and write an XML (RELAX NG?) schema and people use a schema-aware editor to write their stuff.
Cowards? That may be true, but it is irrelevant. I know that in a fist fight, I would probably not win. I'm athletically unfit and not trained to fight (uh huh... that might be why I'm a Slashdot reader). But might does not make right, and any technology that helps make us weedy or fat or unfit guys (and women - hence the thing about rape - men are usually stronger than women, and that's why I'd love to see women be able to defend themselves better) a bit more even in the game of life is good technology, methinks. A gun is just a tool towards that end, and it's one I'd really want beside my bed if I lived in an area where violent attack or burglary was a high risk.
Just as if I have a lot of calculations I would use a computer, if I have a violent attacker coming at me I'd like to have a loaded gun in my hand - it's just a piece of technology that helps solve a problem. The computer helps solve the problem of processing bits of data in series of predefined ways and the gun helps stop humans who are coming towards you with the intent of attacking you. Both can be misused in the wrong hands. But just as governments shouldn't ban the Internet because of child pornography, they shouldn't ban guns on the basis that nasty people will use them but hope that nice people will use them more effectively than the nasty people. A use of an item cannot define that item, however much certain people want it to be so.
Well, perhaps if the cops could actually arrive as a crime is taking place, that might be useful. As might having them on the beat. You still have to pay someone to sit in the little room watching the CCTV cameras - why not pop them out on to the street instead?
Also, if I was living in a south London estate (I don't thankfully, I live down in rural Sussex), I'd rather like to own a gun. The government play this little game where they make 'guns' seem like an evil, evil thing which only south London gangsters would have. I'd like all the respectable, middle class home owners to own guns too. Especially women. Women commit virtually no crime - statistically speaking. If you gave every woman a gun and trained them how to use it, rape would virtually disappear overnight.
How many South London gangstas would be meddling with your house if they knew that the barrel of a shotgun could poke out the letter-box at any moment?
The liberty to not be mugged is tied quite heavily to the liberty to defend yourself using whatever technology you deem appropriate. We do not have that liberty in the UK, and it's a bloody shame that we don't. In the meantime, I'm gonna be keeping a nice, sharp kitchen knife in my bedside table, and you can bet I will defend myself with it. And I'm gonna be looking in to getting a Green Card and going to a state which doesn't see fit to prevent me from defending myself with a gun.
Yes, and that is why the key thing in the Semantic Web space is the making of assertions. If I say that your page is about fishing and you say that your page is about fishing, that's no problem. If you say that your page is about fishing and 90,000 people say it's junk, that's when we have a problem. Having a data model like RDF which allows these assertions to be made across the web is what makes such a thing possible without it all becoming too centralised to be useful.
As the old adage says: if voting did anything, they would have outlawed it already.
Not to poke my nose in to French politics, but it's a perfect example of what I'm saying - both Royal and Sarkozy fit my descriptions of politicians down to a tee. I saw Sarkozy speak last December in Paris and he's a control freak nutbag, a wannabe conservative who'll say all the right-sounding things on crime and immigration but won't actually make our lives less regulated. I imagine Royal isn't much better, although she'll just say all the lefty-sounding things on human rights and welfare but won't actually make us any freer. Both will probably be introducing legislation that'll make any sane Internet user scream and will do nothing to bring technology investment or innovation to Europe. Still, at least YRO will have some fun new stories that it can cover from France if Sarkozy wins.
Elections are once every few years. The polling station is 30 minutes walk or about three minutes drive. It's open from 7am-10pm. The physical process of voting is pretty easy. When I used to live up in the Midlands, it was maybe a ten minute walk to the polling station.
What needs to be fixed is I want someone I can vote for who isn't a self-aggrandizing, liberty-stomping, populist arsebag who will sell out my freedom for whatever trinkets of tabloid approval are currently in vogue. Much as I love the Internet, it is not likely to make our political leaders any less dishonest. They're all fucking useless shits regardless of whether my vote is delivered over dead-tree or TCP/IP.
Plus, I live in one of the most rigidly Conservative-voting constituencies in the country, so my vote is almost totally irrelevant. I'm not sure I'm going to be voting on Thursday.
"owl:sameAs" solves much of that problem - tools like Protégé make it pretty easy to import other ontologies and extend them.
Yes, it's typical "argh! It's not happening!". Well, unless people take an interest and do something, of course it won't happen.
The adoption problem needn't be. If companies and organisations are unwilling to put data up semantically, someone else will. We see this already with accessibility - in the UK, a replacement train times website has been built to replace the crappy National Rail website. I wrote a MySpace screen scraper recently so I wouldn't have to visit the profiles of my friends, and instead I could subscribe to RSS feeds. Semantic data will be available either officially or unofficially quicker than you would think...
So... you let people make their own ontologies. Since the Semantic Web is an open platform with open standards, this is easier than you would think. The only objection to this is "people are stupid". I don't think they are. YMMV.
The Semantic Web layer cake isn't a technology stack per se. There are many ways of implementing each of the pieces in the layer cake.
The following things are what I'd say are currently necessary: URI/IRI, XML and namespaces, the RDF model. Query has been implemented quite neatly with SPARQL but it doesn't require that one have ontologies already in place. Yes, ontology-aware SPARQLing is useful as it solves the Tower of Babel problem to some extent, but it's not necessary.
How we get to Trust does not have to be through Signature and Encryption. For instance, OpenID as a part of the Semantic Web is something I'm very interested in, and I think will play an important part - certainly in the application I'm building.
Dave Beckett puts it brilliantly:
Also, a lot of the parts are fairly immune to breaking. URIs don't 'break' that easily. I can't quite imagine what the alternative would be. The whole thing written in C? Maybe Assembly? Doesn't solve the design goal... :)
That sounds just like a big user experience problem. There is nothing technically wrong with the data model, it just needs to have people design decent, usable interfaces and make the experience of using the system easier. That, combined with motivation - putting incentives in front of people to make data available. Sounds enough to keep the Silicon Valley type startup culture rolling for a few more years... :)
The flipside to this is that the more 'varied' user agents are, the less susceptible they are to exploits. The Semantic Web - broadly understood - could find solutions to SEO. Search personalization and recommendation systems can go some way to neutralize SEO. My inbox on del.icio.us gets no spam at all, because the only people who can post things to it are people I trust. A web based on 'subscriptions' and pubic expressions of 'trust' will be a lot harder to spam and SEO.
Have you seen GRDDL? It's a way of producing RDF/XML data from "plain XML" (although not plain CSV...), and XHTML too (and HTML if you run stuff through Tidy). It's typically W3C - a big long document to explain 'XML+XSLT=RDF', but it's pretty neat nonetheless. Anything that's even vaguely structured can be turned in to RDF very easily. The W3C made a mistake in thinking that RDF etc. would be picked up straight out of the gate in 1999 (when RDF was standardised) - instead, it's taking a rather bending path towards popularity.
There is lots of interesting stuff going on just below the surface, but because it is going on below the surface, it's very easy for people to say there's nothing going on. I have to strongly disagree with Stephen's post. If the MySpaces of the world don't want to expose APIs, that's their prerogative. It'll be uppity programmers who'll write screen scrapers to get them on the semantic web whether they want to be there or not! The market will ebb and flow, and eventually MySpace will be as dead as all the other social networks that sit on the scrapheap of the 'net. Software competes on a lot of things - features being one of them. On the Internet, a site which lets you export your data will be more valuable to the (intelligent) user than one that doesn't. As for the stupid user, well, that's a problem that the Semantic Web can't solve. Stupidity, like lying, is a problem that no amount of W3C standards can solve.
Whitelisting and search personalization. When your search engine is giving priority to people you know and trust and sites they say they trust then the spam problem is significantly lessened. The OpenID community are already using OpenID as a spam whitelisting mechanism. In the Semantic Web layer cake, there is a layer called Trust, which is based on a combination of the RDF stack and document signing and encryption. Even if that isn't the way it's approached, Trust is something that SemWeb people are concerned with. I'm in the middle of designing a SemWebby tool (can't explain it here because I haven't decided whether it's going to be free, commercial or open source etc.) and how trust is dealt with is important.
Currently, Google is the thing to spam - you pick some keywords and you try to spam them. A highly personalized version of Google that uses your friends and colleagues attention data will be a lot harder to spam. With the Semantic Web, yes, people will lie. But if people find out they lie, they can put them on a "liar's list" - just as you do mentally in real life. People lie in databases, people lie on their blogs, people lie in real life. If you expect the Semantic Web to solve a common trait that is present in a large chunk of humanity, then you need to rethink your assumptions about the Semantic Web. :)
Mod parent up. I spent just under a year in the art school at DMU. What a tremendous waste of time. The technical skills I could have taught myself for a lot less than it cost to there. I have a lot of respect for the old concept of polytechnics and vocational education. But these places try too hard to go in to pure science and humanities, and the institutions that come out of the process are a joke. Similarly, the government are trying to implement "professional studies" modules in to more and more of the curricula. Fortunately, we haven't yet had "professional studies" in the philosophy departments, but I can't see it being too far off. British universities get more and more ridiculous by the day.
"funny" is not even half of it.
How about the people who are at the other end talking to this person trying to persuade them that, yes, I have tried plugging it in and it still doesn't work, and that, yes, you should bump me up to someone who actually knows what they are talking about...?
Oh, what we pay our 49 pennies a minute for...
I've gotta agree with some of the presuppositions and points in this review. PHP - in the right hands - is a powerful language. It's great that you can whip things up quickly. But too many of the books go through the fairly simple bits in mind-numbing detail (like, I know what an array is). Thanks to Eclipse and oXygen, most of the time that I'm at a screen, my editor can load the documentation in to a panel while I'm typing it (along with remembering class names, variable names and so on). If I'm coding PHP (or a lot of other languages, for that matter), Eclipse has the language reference. And oXygen gives me the documentation from XSD/RNG/DTD schemas for XML/XHTML etc. Language references aren't useful in dead tree format.
But something that dead tree can be useful for is conveying development experience. Of course, this can be transmitted in other means. Books that give me best practice guidelines are often far more useful than language guides. For instance, in PHP, there is a function called file_get_contents(). It does what it says. You give it a URL or file and it reads it in to a string. But what the language reference *doesn't* tell you is that for getting things off the 'net libcurl is a better way of doing it - it's quicker, more powerful and has a lot of extremely useful options - in short, something which, if one is intent on building a serious web application in PHP, one should probably use. This is one of the reasons why I think language references would be better if managed on a wiki - or, as PHP does it, with comments attached. That way, people can post code samples, bug reports, workarounds, common errors and so on. This is useful.
A measure of a successful technology book these days has to be "does this make me a better developer?". The fact that we have books which deal with best practice means that online documentation has gotten better and better and publishers are responding to that. Most of the languages and frameworks I use I carry the specifications for on my Palm Pilot in Plucker format. Reference books can't compete with that. A few publishers I've seen are shifting towards a tutorial style (in the web design sphere, Friends of Ed is a good example of this).
There are other reasons why web apps tend to be useful:
1. Access from multiple machines (there are some protocols that can handle this reasonably well - IMAP ad Jabber, for instance - but the Web does it without fuss).
2. Open APIs to allow people to reuse existing data in new ways (eg. taking the Chicago crime statistics or the London Tube data and plotting them on a Google Map).
3. Upgrade cycle can be a lot easier - an upgrade can be as easy as "put class.main.php" or "svn commit" etc. Bug-fixing can be done in a similar manner.
That said, having applications that access remote servers is not "just recreating HTML/ Javascript". There's a reason why Amazon has an API to search their products and perform shopping-basket applications - it's so that developers can implement Amazon functionality in to applications - web or otherwise - that could benefit from it. I've been building a web application to allow people to explore each other's music collections. Here, Amazon's data is extremely useful - as I can point people from my application to go to Amazon and buy the CD. There are even classes available in Perl (DBD::Amazon and various others - search for 'Amazon' modules on CPAN), PHP (Services_Amazon via PEAR) and Python (Google for 'pyamazon', I think).
You say that it would be bad for each web site to have to create a desktop application. That's not where things are going. What needs to happen is for each of the companies to release a sensible, sane API, offer some motivation (affiliate programme is one way), and the apps will follow.
Do you think my parents are going to be telnetting in to their mail server? Seriously.
One of the good things about Ajax is that it's asynchronous, which means that it can work on unreliable connections reasonably well. I can open up Gmail on my laptop while I'm at home on my DSL connection. Then I can shut my laptop, open it on the commuter train where my connection is an unreliable mobile phone modem connection and, though the service may come in and out, I don't need to log in again. I use SSH a lot, and it's almost impossible to use over an unreliable modem connection. Most of the better programmed Ajax apps run fine though.
If you want to replicate this experience, grab yourself a GPRS connection (I dunno what they call that over in America, but it's not EVDO, it's the slow one that is used on the Treo 650 etc.) and try logging in to a telnet or SSH connection.
There *are* promising developments to solve this problem. One is Adobe Flex, which will allow developers to release simple web applications for the desktop that sync up with online servers. Another is XForms, an XML specification for forms, which could be combined with, say, a Java application to make very simple GUI applications. The final way is just to wait. Connectivity will be everywhere eventually. Even railway tunnels.
Please come and work at my university. We need sane systems people. We have a Windows + wi-fi setup. It sucks. It's slow. The filters are extremely badly configured - the letters "mp3" are filtered from all URLs (no distinction between 'download a Jay-Z track' and 'download an MP3 of an academic debate'). The deskop environment is locked down so that the only applications that a user can use are Word and Internet Explorer. I wanted to work on an ASCII file the other day and Notepad is filtered. I shit you not.
I'd make a big fuss about it, but I finish in a couple of months and most of the time I can get around the filtering (SSH in to web server, wget/links - might set up a tunnel if I get bored).
s/middle/management or s/middle/middle management. And damn this absent-minded brain of mine running on auto-idiot mode.
How about "so easy, even middle can do it"? Or "so easy, even George Bush could do it".
:)
That said, having spent a few days trying to get Java to install properly on my Ubuntu machine so that I can run oXygenXML, it's *not* easy. Multiple versions not working with one another, different dependencies. It should be pre-installed, or at the very most "sudo apt-get install java". Perhaps I'm better suited to a management position...
Yes, it's a masterful move on the part of Microsoft's PR department.
If you want to satisfy large government and corporate buyers who want the data security that XML/(your favourite open data interchange standard) brings, you have three options:
1. Make a sensible, well-designed XML standard to facilitate interop.
2. Don't bother.
3. Make an insanely complex XML standard to facilitate interop in theory while hindering it in practice.
Microsoft have chosen the last option. Anybody surprised?