"Hello, Vendor. We've spent several man months trying to get your software to work, but we still get random lock-ups. What do you suggest?"
"Hi, customer. Could you please spend several man-weeks running these complex diagnostic tests on our software, so we can try to fix it. Yes - very similar to the last set of diagnostic tests - yes, the ones that didn't help us diagnose the problem - yup, like the ones before that too, but this time with a few different settings - yes, please, if you could send us all that data then we'll have a bit more time to think up some excuses. Thanks!"
Now, I've no idea what the particulars here are, but I've been in plenty of situations where it's a waste of time sending vendor support yet more giant error logs, and running yet more diagnostic tests. I used to regularly 'help' Vignette try to fix our problems, but they never did. I could have spend the time better doing our own workarounds - or switching CMS vendors, which is what we ended up doing.
I think you are missing the point. Getting your software to build on two platforms is about 10% of the total problem.
Think of all the surrounding stuff: 1. Manuals 2. Installation 3. Interfaces to other parts of the OS
So what if your code compiles on Solaris and Linux. If you want to support both, you will need to write a Solaris package and an RPM package. And one system uses/bin/sh as the default user shell and the other user/bin/bash. And those two shells don't work the same way. And a solaris user might well expect the program to be installed in/opt, while the linux sysadmin might well want it in/usr/local. And what if the program relies on the system cron to schedule things. You think Linux and Solaris cron work _exactly_ the same way?
Never mind that digital audio broadcasting is not significantly greater in quality than regular, analog radio
Well, not in the UK. Digital Audio Broadcast (aka digital radio) has much better quality than FM - and that's assuming you can get good FM reception, which is rare here.
No, actually it is the same. Ask anyone on the street what "I like eating cheeseburgers more than her" means, and the vast majority will not think of cannibalism.
Well duh, of course not. They'll be thinking of oral sex like everyone else.
Well, maybe it was like that and maybe it wasn't. I've never been in combat, and with a bit of luck I won't be. But friends and relatives of mine have been, and they certainly haven't described it permanent terrifying chaos. Moments of that, yes, but not all of it.
The point is that we hope to train soldiers to operate despite the fear. We train paramedics to deal with train wrecks. We recognise that they are stressful unpleasant situations, but we don't accept the excuse that there were lots of screaming people and burnt corpses as a reason for a paramedic administering fatal doses of anaesthetic by mistake.
Likewise, we should not accept the fog of war as an excuse for shooting your own side. It's never good enough to simply say "It's war, shit happens". That's like saying after every plane crash "Hey, it's hundreds of tons of metal floating in mid air - they can't all stay up!"
HTML is NOT the universal data format over HTTP. By far the most data transfered over HTTP is MIME encoded binary data or one kind or another (mostly gifs, jpgs, mp3s etc.
In case you hadn't noticed HTML is a subset of text. In case you hadn't noticed ASCII is a piece of crap that should have died years ago - or are you suggesting that it's innapropriate for Japanese people to communicate via SMTP email?
It's perfectly fine to use SMTP to send Unicode text data. Why is it not fine to use SMPT to send HTML text data? Why would it be a feature to prevent an email message having embedded images? Do you think it's bad the way some word processors can embed spreadsheets? Do you think we should force the separation into different applications "where they belong"?
Do you think its bad that email clients support hyperlinking from plain text emails, on the grounds that hyperlinking belongs to the web, and hey, this is _email_? That would be dumb, right?
Do you also object to web based email clients, on the grounds that hey - this is the web - cut out the email!?
The sooner geeks get over the fact that technology moves on the better.
Marketing emails (whether spam or legit) are always _much_ more effective if they are HTML. Maybe not with you, but you aren't important. If you want the world at large to get a message, sending it in a (well designed) HTML email *is* more effective. It's that simple. Deal with it.
The model you describe is interesting, but it is not universal. Are you describing a computing related journal?
The publisher I work for does all of the jobs you describe in house (except for peer review), and they are all full time positions, funded by the publisher. We would never dream of expecting an author to know what LaTeX was, let alone submit in it. In addition, we perform several other steps (technical editing, statistics checking, etc) that are all costs to us.
So, I'm willing to believe that some publishers are riding on their name alone. They do little work beyond organising peer review and simplistic typsetting and printing, but can charge enormous prices because all the best research is in their journal.
But, there are many high quality, expensive publications, that are working hard for their money.
"My personal amazement never diminishes when companies or institutions complain over a competitor offering a good/service at lesser or no cost to the consumer."
Errr. So you expect companies to be happy when they suddenly face stiff competition? Why wouldn't they complain, this is obviously bad for them. Now, it might be good for us the consumer, but it's perfectly reasonable for someone like Elsevier to object to it and to try to convince the market that it's bad.
The market will decide. I don't see the problem. Most people are happy to use free software rather than pay more for a high quality program that does a similar thing. Look at IE vs Opera. I am sure in the future we will get easy, cheap, access to slightly lower quality academic papers.
" All the academic publishers do is marketing, printing, type setting, and mailing to libraries, and none of those are essential for academic journals anymore."
I'm getting tired of responding to this. Scientific publishers also:
1. Organise peer review - it's not like it organises itself. You have to find and select peers without conflicting interests, but with adequate subject knowledge. You have to chase them for deadlines and give them help as needed. 2. Edit. In real life, scientists and academics are often not very good at writing cogently. Someone needs to help make it readable. 3. Copy Edit. And someone needs to fix the grammar and spelling. 4. Technical Edit. Someone needs to check all the mathematical formulae, diagrams, drug names, etc. to make sure nothing's wrong. 5. Statistics. In many fields, the publisher will use a statistician to confirm that there are no statistical errors in the paper. 6. Original writing. Most academic journals, and all the major ones include a considerable amount of original content as well as research papers. All that needs to be written and edited.
Now all of this hard work is OPTIONAL. The research has done their experiment, and written it up. They own the copyright to that. They can (and often do) stick that version up on their website. Fine. Now, if someone wants to read that original, and ignore the spelling mistakes, and do their own double-checking of all the figures, and chase up the peer reviewers to make sure they aren't personal friends and so forth then they can get that paper for free.
Or, they can subscribe to a journal that does all that for them, and drops it on their doormat once a month complete with an overview of the state of the art by the editor, an amusing comic, some classified ads, and whatever else the journal wants to add.
I can't think of any journal that today asks authors to give up copyright to papers they submit. It's a choice.
Now, it's also true that the journals are worried. Maybe no-one _does_ want to pay for fixing typos and checking figures? If that's the case, then journals are in trouble. All we have to do is wait and see.
That's an understatement. I read the original description of the timing security flaw, and it was far from practical to exploit. Let's just say that you needed a real-time key logger on your target user, and the ability to type _very_ fast just to get started. I'm not aware of the weakness ever being exploited, and it was fixed prior to 2000.
Meanwhile, people are still coming up with amusing weaknesses! Here's one that merely requires stealing the user's token for a week without their knowledge, and having access to a digital camera, an accurate OCR application, and two months worth of CPU time.
Re:Innovate, not copy
on
Gates on Google
·
· Score: 5, Interesting
Microsoft innovate all the time. Just not particularly in writing software programs. But then not many companies do - the history of software is the history of incremental improvements - no one innovates that much. Google is merely grep version 9082.1 , and even the clever bits of Google were done 'first' in research instituations around the world.
MS's only big software innovation has been integration. They realised that people don't want programs. They want a computer. One thing that does everything in a consistent joined up manner. That _WAS_ innovation. Everyone else at the time still thought it was a _good_ thing to have lots of little programs each with it's own purpose, UI, etc tailored to a specific job.
MS realised that this was crap, and to the annoyance of software people everywhere, MS was right. Most people want to buy a word processor and a spreedsheet from different companise in the same way they want to buy their hob and their oven from different companies. Not at all.
I would also say that ASP pages were innovative - not so much the idea of templates, but the idea of creating a proper web SDK, with a cohesive set of classes. It's not rocket science, but no-one else had thought of offering a complete solution to what was _still_ being viewed as a set of separate problems - a web server, a programming language, a database API, etc. etc.
However, where MS is _really_ innovative is in marketing. They have found ways to promote and market software that no-one else has ever thought of. Now, those ways may not be 'nice' but they are certainly innovative.
But yes, you can convert WordML into other XML based formats if you are very good at XSLT and very patient.
Alternatively, get the professional version of Word and get it to save documents in XML that complies with your own schema. Costs a good deal more, but is rather a good feature.
Yup, that's software quality alright. I mean, look at windows for workgroups 3.11, and compare it with crap like XP or 2000 - we've lost so much stability, and performance, these modern OSes are just rubbish compared with the old ones. Don't get me started on how bad OSX is!
Another bit of software that's been getting worse is Photoshop. I mean, have you ever tried using version 1? You can do _so much_ more than you can with the current version. They just keep removing features with each new release, and the software gets worse!
It's the same with databases. It used to be that everything used fixed length fields, and really restrictive character sets. That meant that people like Mr Rénauld-Smythe could rely on always being refered to as Mr Renauldsmyth by their gas company. Nowadays, that kind of attention to detail and users is completely absent.
And it's not just in ways like this that software quality is going down-hill. Customer services is going to the dogs! I remember when, if I wanted an update to my software, I could write a letter, then wait for a week to get some floppy disks with a patch on. Nowadays I have to connect to some huge wide area high speed network and download the patches myself! Just because the software companies want to save the cost of postage! Well I ask you.
In every way, from speed, features, stability and customer service, software is getting worse and worse. I was so glad when Open Source came along and changed it! No sooner had Microsoft scrapped the excellent Windows 3.1 environment, and replaced it with the dreadfull Win95 one, but Linux came along with - X11 and twm! I thought quality and useability like that was dead!
And that's not all. I remember when configuring a PC let you insert your own IRQ numbers and decide what drivers were loaded into what RAM segments - and then, DUH, Microsft figured they should do all that for us - as if we weren't clever enough to resolve hardware addressing issues ourselves! Imagine my delight when I found Linux. I spent _many_ happy hours manually configuring my drivers, I can tell you! That's the kind of quality I wanted.
From the simplicity and ease of LaTeX, to the high performance and slick modernity of X11, there's nothing that OSS hasn't done better than their so-called rivals. It's true that some things are getting worse - ReiserFS instead of Ext2? I don't think so! But the for most important things, like printer configuration, and having a fully skinable CD player applet with it's own LISP based configuration language - well, Open Source is way out in front.
P.S. I was disappointed to see that Opera is making such poor software - that's why I'm sticking to Netscape 2.1
0. Not ditching the workstation market soon enough. Sun used to make lots of workstations. Anyone on the outside could see Intel/NT was going to eat these up in seconds. Sun held on far too long, although this didn't wipe them out like it did SGI.
1. Profiteering from the demise of HP/IBM/SGI. For a period of about 10 years, Solaris on Sparc was pretty much the only safe solution for many large organisations. Sun realised this and gouged customers pretty heavily. This made senior IT people hate them, and accelerated the move to NT.
2. Ditching Solaris on Intel. Sun used to make a free as in beer distro of Solaris to run on Intel. It had limited hardware support, but it was easy enough to get going on the machines from the big vendors. This was an excellent way of getting Solaris used for things that might have been shifted to NT or Linux, but for some stupid reason they just cut support for it on day. Pissed off huge numbers of customers. I think the idea was they were introducing new low-end sparc machines that were going to be 'as cheap as Intel servers'. Yeah, right - let's compete with Dell, that's work.
3. Java. I don't think Sun has made much money from Java, and it's been a huge distraction. Are Sun the solid trusted makers of server hardware and server OSes, or are they funky cool bleeding edge software guys changing the face of the internet? Rather obviously the former, but disasterous crap like JavaStations and so on took up a huge amount of time and effort.
Sun should have made Java an open specification like, err, EVERY OTHER FRIGGING LANGUAGE EVER MADE, instead of fighting idiotic lawsuits with MS (who were in the right for a change).
You appear not to have done very much with Solaris. It seems pretty clear that Linux cannot support very sophisticated hardware (say, more than 8 CPUs), whereas Solaris can.
Even ignoring that, having used both OSes, I prefer Solaris. Linux distribution chaos more or less removes the advantage of what 'commercial' support there is for Linux. Thus, you can't say "Sybase runs on Linux". You can only say "Sybase runs on some particular versions of some particular distributions of Linux" - and those versions are ones that contain plenty of proprietary (non GPL) code.
Now, even if we are ignoring commercial apps, and just, say, running Apache, Samba, and other solid OSS apps. Well, all these work fine on Solaris. But Solaris isn't shovelware. Solaris doesn't try to install Postgres or Mysql as part of the default application set. Solaris doesn't require a graphics card. Solaris doesn't install GNU's halfwitted 'info' pages. Solaris has better documentation that Linux. Solaris doesn't have insane library compatibility conflicts between different versions of libgc libc libstdc++, gcc, etc.
The only think going for Linux is its freeness, its support of cheap, fast, intel hardware, and the fact that many OSS projects are now a complete pain to compile and install on anything other than Linux.
It amuses me that Linux will end up being an inferior OS with near-universal deployment, that remains in place because of the large number of applications that can only be run on it. Plus ca change...
Unfortunately not. Ramping up CPU production takes a very long time, because you have to build big, complex, factories. Ramping up demand can happen extremely quickly, because it just requires someone selecting the 'amd' dropdown rather than the 'intel' one on the Dell site.
That is (one reason why) Dell isn't going AMD.
For those who hadn't noticed (all geeks, apparently) Dell is not a computer company. It is a manufacturing company. It has made great, innovative strides in extremely efficient manufacturing. They just happen to be doing it with computers. The choice of CPU in the box means as much to Dell as the choice of compressor in a Siemens fridge means to Siemens. And the CPU means as much to most consumers as the compressor in their fridge.
Sure, some compressors are probably more efficient, reliable, quiet, whatever. But if you have n million invested in robots that can install one brand of compressor and not another, are you going to switch? What if you have a huge set of automated hardware tests all of them based on Intel motherboards - are you going to switch? What if you have hundreds of call center scripts based around diagnosing known Intel related problems?
The actual capability of the processor is NOT RELEVANT. Dell don't sell PCs because they make the best ones. They sell them because they make them very cheap, and can deliver huge orders on time, even when those orders are semi-custom. Dell does not achieve this impressive feat by switching components every time one of them becomes slightly better than another. If you want that kind of thing, there are many PC makers who will do it.
I work for a major medical publisher, so I'll explain what I can...
"I never understood the economics of peer-reviewed scientific journals. The authors don't get any money"
This is correct, authors are not paid for submissions.
" and are usually tech-savvy enough to produce well-formatted papers."
This is incorrect. Formatting even simple papers is difficult, let alone ones with complex graphs and tables. It's not something an author can (or wants) to do. In our case we mark the data up in a complex XML schema and do some clever layout things to format the articles. Many places just do it by hand in Quark or whatever.
" The peer-reviewers (at least when I peer reviewed) didn't get any money."
Correct.
" And being an editor is an academic feather in your cap."
Incorrect. Articles are edited by professional editors. That means you need, at _least_, a doctorate in medicine, a very high standard of English, and several years editorial experience. This isn't a cheap person to employ, and you need many of them. These people are of course helped by a team of professional sub-editors and copy-editors.
"So the cost of content and the cost of reviewing the content is close to zero."
Not at all. The other major cost is the cost of reviewing papers. A major journal will receive ten times more papers than it can publish. Each one needs to be read and evaluated. They must _all_ be read by _several_ people with the knowledge to actually understand what the paper is talking about. Then those people must meet weekly (or however often the publication comes out) and decide which papers are in and which are out. Then the whole peer review and editorial process begins.
Other jobs that cost money are:
Statisticians. A professional is needed to check the figures and calculations in the papers, as they are often wrong.
Production assistants. Peer reviewers and authors are not paid. This gives them little incentive to do things either on time or in the way they are asked. Someone has to nicely chase them and organise them and help them.
Technical people (like me!). Converting large amounts of complex XML into things like printer ready PDF (that's as in commercial printer, not laserjet), XHTML, exports for pubmed etc, is not trivial.
" But some journals cost individuals and especially the institutions a large amount of money. In this day of electronic typesetting and distribution, does it make any sense?"
Yes. Electronic typesetting is not cheap, not is something automatic just because its electronic. A high quality journal cannot be laid out by machine. A human has to decide where articles go, how figures are positions etc. No layout engine we've ever seen is up to this except in simple cases.
Take the New England Journal of Medicine. It's about $150 for an individual subscription and ranges from $1000 to $17,000 for institutions depending on the size. This is for a publication that doesn't pay authors, and in fact can make authors bend over backwards. No wonder all sorts of publication models are being explored.
Messing with the user agent string won't work, as googlebots use a well known IP range, and you can ID google like this. It's how we do, and of course it protects you from people _pretending_ to be googlebot but changing their UA string...
The other problem is that many legit sites, including my own, use 302 redirects a lot. They are used a great deal for maintaining backward compatability with old links when URL schemes change. They are also very useful in authentication and personalisation siutations
I don't get it. So they send these threatening letters to a few people - the so-called 'major abusers' or whatever, in the hope that the hundreds of thousands of 'home abusers' who just copy a few lines or, share their favourite algorithms with personal friends, are really going to change.
Frankly, if the GNU crowd want to stop copyright abuse, they'd better start outputting some source code actually worth protecting. I mean, have you _looked_ at sourceforge recently? It's all half finished projects and crappy PHP webapps. Who cares if we abuse the copyright on that stuff?
Basically, the new GNU license abuse is here to stay, and the Free Software industry better get used to it. Instead, they were lost in their own self-congratulatory world, talking up a few chart-toppers like Apache and Tomcat, while all around them hundrends of little start-up software companies were ripping the source code from apps and embedding in commercial stuff.
Basically, the FSF needs to ditch their bully boy tactics, and deal with the fact that copy-and-paste theft of neat bits of GNU code are here to stay. Sure, go imprison some college kids, if you want to stoop that low. But I for one will still be using peer to peer networks and the next generation of anonymous file sharing networks to obtain Free Software source code, and compile into my applications without telling anyone.
"May be it is because Microsoft donates money and its software to the schools at subsidized prices that not many students learn about Linux?"
Oh, yes, that'll be it. Who'd learn about foo for free when they can get bar for almost free? I mean, the main reason Christianity is so popular in the US is because bibles are heavily subsidised.
Meanwhile, in reality land, it could be that Indian programmers would like maximise their chances of gainful employment, and so learn to use the most widespread OS? Gee, I wonder if that could be it.
Jon 'try the horse in front of the cart next time' Peterson.
Slashdot Mirror
"Hello, Vendor. We've spent several man months trying to get your software to work, but we still get random lock-ups. What do you suggest?"
"Hi, customer. Could you please spend several man-weeks running these complex diagnostic tests on our software, so we can try to fix it. Yes - very similar to the last set of diagnostic tests - yes, the ones that didn't help us diagnose the problem - yup, like the ones before that too, but this time with a few different settings - yes, please, if you could send us all that data then we'll have a bit more time to think up some excuses. Thanks!"
Now, I've no idea what the particulars here are, but I've been in plenty of situations where it's a waste of time sending vendor support yet more giant error logs, and running yet more diagnostic tests. I used to regularly 'help' Vignette try to fix our problems, but they never did. I could have spend the time better doing our own workarounds - or switching CMS vendors, which is what we ended up doing.
Brass??!?!?
It's bronze! Bronze age! Can you imagine people going around trying to use brass tipped spears?
I think you are missing the point. Getting your software to build on two platforms is about 10% of the total problem.
/bin/sh as the default user shell and the other user /bin/bash. And those two shells don't work the same way. And a solaris user might well expect the program to be installed in /opt, while the linux sysadmin might well want it in /usr/local. And what if the program relies on the system cron to schedule things. You think Linux and Solaris cron work _exactly_ the same way?
Think of all the surrounding stuff:
1. Manuals
2. Installation
3. Interfaces to other parts of the OS
So what if your code compiles on Solaris and Linux. If you want to support both, you will need to write a Solaris package and an RPM package. And one system uses
It's not straightforward.
Well, not in the UK. Digital Audio Broadcast (aka digital radio) has much better quality than FM - and that's assuming you can get good FM reception, which is rare here.
Well duh, of course not. They'll be thinking of oral sex like everyone else.
"The web flattens time by making more of the past accessible."
Wow, not on my version of Firefox it doesn't. I gotta upgrade!
Well, maybe it was like that and maybe it wasn't. I've never been in combat, and with a bit of luck I won't be. But friends and relatives of mine have been, and they certainly haven't described it permanent terrifying chaos. Moments of that, yes, but not all of it.
The point is that we hope to train soldiers to operate despite the fear. We train paramedics to deal with train wrecks. We recognise that they are stressful unpleasant situations, but we don't accept the excuse that there were lots of screaming people and burnt corpses as a reason for a paramedic administering fatal doses of anaesthetic by mistake.
Likewise, we should not accept the fog of war as an excuse for shooting your own side. It's never good enough to simply say "It's war, shit happens". That's like saying after every plane crash "Hey, it's hundreds of tons of metal floating in mid air - they can't all stay up!"
HTML is NOT the universal data format over HTTP. By far the most data transfered over HTTP is MIME encoded binary data or one kind or another (mostly gifs, jpgs, mp3s etc.
In case you hadn't noticed HTML is a subset of text. In case you hadn't noticed ASCII is a piece of crap that should have died years ago - or are you suggesting that it's innapropriate for Japanese people to communicate via SMTP email?
It's perfectly fine to use SMTP to send Unicode text data. Why is it not fine to use SMPT to send HTML text data? Why would it be a feature to prevent an email message having embedded images? Do you think it's bad the way some word processors can embed spreadsheets? Do you think we should force the separation into different applications "where they belong"?
Do you think its bad that email clients support hyperlinking from plain text emails, on the grounds that hyperlinking belongs to the web, and hey, this is _email_? That would be dumb, right?
Do you also object to web based email clients, on the grounds that hey - this is the web - cut out the email!?
The sooner geeks get over the fact that technology moves on the better.
Marketing emails (whether spam or legit) are always _much_ more effective if they are HTML. Maybe not with you, but you aren't important. If you want the world at large to get a message, sending it in a (well designed) HTML email *is* more effective. It's that simple. Deal with it.
The model you describe is interesting, but it is not universal. Are you describing a computing related journal?
The publisher I work for does all of the jobs you describe in house (except for peer review), and they are all full time positions, funded by the publisher. We would never dream of expecting an author to know what LaTeX was, let alone submit in it. In addition, we perform several other steps (technical editing, statistics checking, etc) that are all costs to us.
So, I'm willing to believe that some publishers are riding on their name alone. They do little work beyond organising peer review and simplistic typsetting and printing, but can charge enormous prices because all the best research is in their journal.
But, there are many high quality, expensive publications, that are working hard for their money.
"My personal amazement never diminishes when companies or institutions complain over a competitor offering a good/service at lesser or no cost to the consumer."
Errr. So you expect companies to be happy when they suddenly face stiff competition? Why wouldn't they complain, this is obviously bad for them. Now, it might be good for us the consumer, but it's perfectly reasonable for someone like Elsevier to object to it and to try to convince the market that it's bad.
The market will decide. I don't see the problem. Most people are happy to use free software rather than pay more for a high quality program that does a similar thing. Look at IE vs Opera. I am sure in the future we will get easy, cheap, access to slightly lower quality academic papers.
" All the academic publishers do is marketing, printing, type setting, and mailing to libraries, and none of those are essential for academic journals anymore."
I'm getting tired of responding to this. Scientific publishers also:
1. Organise peer review - it's not like it organises itself. You have to find and select peers without conflicting interests, but with adequate subject knowledge. You have to chase them for deadlines and give them help as needed.
2. Edit. In real life, scientists and academics are often not very good at writing cogently. Someone needs to help make it readable.
3. Copy Edit. And someone needs to fix the grammar and spelling.
4. Technical Edit. Someone needs to check all the mathematical formulae, diagrams, drug names, etc. to make sure nothing's wrong.
5. Statistics. In many fields, the publisher will use a statistician to confirm that there are no statistical errors in the paper.
6. Original writing. Most academic journals, and all the major ones include a considerable amount of original content as well as research papers. All that needs to be written and edited.
Now all of this hard work is OPTIONAL. The research has done their experiment, and written it up. They own the copyright to that. They can (and often do) stick that version up on their website. Fine. Now, if someone wants to read that original, and ignore the spelling mistakes, and do their own double-checking of all the figures, and chase up the peer reviewers to make sure they aren't personal friends and so forth then they can get that paper for free.
Or, they can subscribe to a journal that does all that for them, and drops it on their doormat once a month complete with an overview of the state of the art by the editor, an amusing comic, some classified ads, and whatever else the journal wants to add.
I can't think of any journal that today asks authors to give up copyright to papers they submit. It's a choice.
Now, it's also true that the journals are worried. Maybe no-one _does_ want to pay for fixing typos and checking figures? If that's the case, then journals are in trouble. All we have to do is wait and see.
That's an understatement. I read the original description of the timing security flaw, and it was far from practical to exploit. Let's just say that you needed a real-time key logger on your target user, and the ability to type _very_ fast just to get started. I'm not aware of the weakness ever being exploited, and it was fixed prior to 2000.
l letin%2003-002
Meanwhile, people are still coming up with amusing weaknesses! Here's one that merely requires stealing the user's token for a week without their knowledge, and having access to a digital camera, an accurate OCR application, and two months worth of CPU time.
http://www.okiok.com/index.jsp?page=Security%20Bu
Microsoft innovate all the time. Just not particularly in writing software programs. But then not many companies do - the history of software is the history of incremental improvements - no one innovates that much. Google is merely grep version 9082.1 , and even the clever bits of Google were done 'first' in research instituations around the world.
MS's only big software innovation has been integration. They realised that people don't want programs. They want a computer. One thing that does everything in a consistent joined up manner. That _WAS_ innovation. Everyone else at the time still thought it was a _good_ thing to have lots of little programs each with it's own purpose, UI, etc tailored to a specific job.
MS realised that this was crap, and to the annoyance of software people everywhere, MS was right. Most people want to buy a word processor and a spreedsheet from different companise in the same way they want to buy their hob and their oven from different companies. Not at all.
I would also say that ASP pages were innovative - not so much the idea of templates, but the idea of creating a proper web SDK, with a cohesive set of classes. It's not rocket science, but no-one else had thought of offering a complete solution to what was _still_ being viewed as a set of separate problems - a web server, a programming language, a database API, etc. etc.
However, where MS is _really_ innovative is in marketing. They have found ways to promote and market software that no-one else has ever thought of. Now, those ways may not be 'nice' but they are certainly innovative.
It's the being around for a long time that I worry about.
HA ha ha ha. Ohh, my that's funny.
WordML and 'simple' don't really go together.
But yes, you can convert WordML into other XML based formats if you are very good at XSLT and very patient.
Alternatively, get the professional version of Word and get it to save documents in XML that complies with your own schema. Costs a good deal more, but is rather a good feature.
Yup, that's software quality alright. I mean, look at windows for workgroups 3.11, and compare it with crap like XP or 2000 - we've lost so much stability, and performance, these modern OSes are just rubbish compared with the old ones. Don't get me started on how bad OSX is!
Another bit of software that's been getting worse is Photoshop. I mean, have you ever tried using version 1? You can do _so much_ more than you can with the current version. They just keep removing features with each new release, and the software gets worse!
It's the same with databases. It used to be that everything used fixed length fields, and really restrictive character sets. That meant that people like Mr Rénauld-Smythe could rely on always being refered to as Mr Renauldsmyth by their gas company. Nowadays, that kind of attention to detail and users is completely absent.
And it's not just in ways like this that software quality is going down-hill. Customer services is going to the dogs! I remember when, if I wanted an update to my software, I could write a letter, then wait for a week to get some floppy disks with a patch on. Nowadays I have to connect to some huge wide area high speed network and download the patches myself! Just because the software companies want to save the cost of postage! Well I ask you.
In every way, from speed, features, stability and customer service, software is getting worse and worse. I was so glad when Open Source came along and changed it! No sooner had Microsoft scrapped the excellent Windows 3.1 environment, and replaced it with the dreadfull Win95 one, but Linux came along with - X11 and twm! I thought quality and useability like that was dead!
And that's not all. I remember when configuring a PC let you insert your own IRQ numbers and decide what drivers were loaded into what RAM segments - and then, DUH, Microsft figured they should do all that for us - as if we weren't clever enough to resolve hardware addressing issues ourselves! Imagine my delight when I found Linux. I spent _many_ happy hours manually configuring my drivers, I can tell you! That's the kind of quality I wanted.
From the simplicity and ease of LaTeX, to the high performance and slick modernity of X11, there's nothing that OSS hasn't done better than their so-called rivals. It's true that some things are getting worse - ReiserFS instead of Ext2? I don't think so! But the for most important things, like printer configuration, and having a fully skinable CD player applet with it's own LISP based configuration language - well, Open Source is way out in front.
P.S. I was disappointed to see that Opera is making such poor software - that's why I'm sticking to Netscape 2.1
Sun made a few huge mistakes:
0. Not ditching the workstation market soon enough. Sun used to make lots of workstations. Anyone on the outside could see Intel/NT was going to eat these up in seconds. Sun held on far too long, although this didn't wipe them out like it did SGI.
1. Profiteering from the demise of HP/IBM/SGI. For a period of about 10 years, Solaris on Sparc was pretty much the only safe solution for many large organisations. Sun realised this and gouged customers pretty heavily. This made senior IT people hate them, and accelerated the move to NT.
2. Ditching Solaris on Intel. Sun used to make a free as in beer distro of Solaris to run on Intel. It had limited hardware support, but it was easy enough to get going on the machines from the big vendors. This was an excellent way of getting Solaris used for things that might have been shifted to NT or Linux, but for some stupid reason they just cut support for it on day. Pissed off huge numbers of customers. I think the idea was they were introducing new low-end sparc machines that were going to be 'as cheap as Intel servers'. Yeah, right - let's compete with Dell, that's work.
3. Java. I don't think Sun has made much money from Java, and it's been a huge distraction. Are Sun the solid trusted makers of server hardware and server OSes, or are they funky cool bleeding edge software guys changing the face of the internet? Rather obviously the former, but disasterous crap like JavaStations and so on took up a huge amount of time and effort.
Sun should have made Java an open specification like, err, EVERY OTHER FRIGGING LANGUAGE EVER MADE, instead of fighting idiotic lawsuits with MS (who were in the right for a change).
Errr...
You appear not to have done very much with Solaris. It seems pretty clear that Linux cannot support very sophisticated hardware (say, more than 8 CPUs), whereas Solaris can.
Even ignoring that, having used both OSes, I prefer Solaris. Linux distribution chaos more or less removes the advantage of what 'commercial' support there is for Linux. Thus, you can't say "Sybase runs on Linux". You can only say "Sybase runs on some particular versions of some particular distributions of Linux" - and those versions are ones that contain plenty of proprietary (non GPL) code.
Now, even if we are ignoring commercial apps, and just, say, running Apache, Samba, and other solid OSS apps. Well, all these work fine on Solaris. But Solaris isn't shovelware. Solaris doesn't try to install Postgres or Mysql as part of the default application set. Solaris doesn't require a graphics card. Solaris doesn't install GNU's halfwitted 'info' pages. Solaris has better documentation that Linux. Solaris doesn't have insane library compatibility conflicts between different versions of libgc libc libstdc++, gcc, etc.
The only think going for Linux is its freeness, its support of cheap, fast, intel hardware, and the fact that many OSS projects are now a complete pain to compile and install on anything other than Linux.
It amuses me that Linux will end up being an inferior OS with near-universal deployment, that remains in place because of the large number of applications that can only be run on it. Plus ca change...
Unfortunately not. Ramping up CPU production takes a very long time, because you have to build big, complex, factories. Ramping up demand can happen extremely quickly, because it just requires someone selecting the 'amd' dropdown rather than the 'intel' one on the Dell site.
That is (one reason why) Dell isn't going AMD.
For those who hadn't noticed (all geeks, apparently) Dell is not a computer company. It is a manufacturing company. It has made great, innovative strides in extremely efficient manufacturing. They just happen to be doing it with computers. The choice of CPU in the box means as much to Dell as the choice of compressor in a Siemens fridge means to Siemens. And the CPU means as much to most consumers as the compressor in their fridge.
Sure, some compressors are probably more efficient, reliable, quiet, whatever. But if you have n million invested in robots that can install one brand of compressor and not another, are you going to switch? What if you have a huge set of automated hardware tests all of them based on Intel motherboards - are you going to switch? What if you have hundreds of call center scripts based around diagnosing known Intel related problems?
The actual capability of the processor is NOT RELEVANT. Dell don't sell PCs because they make the best ones. They sell them because they make them very cheap, and can deliver huge orders on time, even when those orders are semi-custom. Dell does not achieve this impressive feat by switching components every time one of them becomes slightly better than another. If you want that kind of thing, there are many PC makers who will do it.
WTF? You mean the people who gave the made the Gimp skinnable were _specialists_ in UI design? I'd sure hate to see what amateurs would have done.
I work for a major medical publisher, so I'll explain what I can...
"I never understood the economics of peer-reviewed scientific journals. The authors don't get any money"
This is correct, authors are not paid for submissions.
" and are usually tech-savvy enough to produce well-formatted papers."
This is incorrect. Formatting even simple papers is difficult, let alone ones with complex graphs and tables. It's not something an author can (or wants) to do. In our case we mark the data up in a complex XML schema and do some clever layout things to format the articles. Many places just do it by hand in Quark or whatever.
" The peer-reviewers (at least when I peer reviewed) didn't get any money."
Correct.
" And being an editor is an academic feather in your cap."
Incorrect. Articles are edited by professional editors. That means you need, at _least_, a doctorate in medicine, a very high standard of English, and several years editorial experience. This isn't a cheap person to employ, and you need many of them. These people are of course helped by a team of professional sub-editors and copy-editors.
"So the cost of content and the cost of reviewing the content is close to zero."
Not at all. The other major cost is the cost of reviewing papers. A major journal will receive ten times more papers than it can publish. Each one needs to be read and evaluated. They must _all_ be read by _several_ people with the knowledge to actually understand what the paper is talking about. Then those people must meet weekly (or however often the publication comes out) and decide which papers are in and which are out. Then the whole peer review and editorial process begins.
Other jobs that cost money are:
Statisticians. A professional is needed to check the figures and calculations in the papers, as they are often wrong.
Production assistants. Peer reviewers and authors are not paid. This gives them little incentive to do things either on time or in the way they are asked. Someone has to nicely chase them and organise them and help them.
Technical people (like me!). Converting large amounts of complex XML into things like printer ready PDF (that's as in commercial printer, not laserjet), XHTML, exports for pubmed etc, is not trivial.
" But some journals cost individuals and especially the institutions a large amount of money. In this day of electronic typesetting and distribution, does it make any sense?"
Yes. Electronic typesetting is not cheap, not is something automatic just because its electronic. A high quality journal cannot be laid out by machine. A human has to decide where articles go, how figures are positions etc. No layout engine we've ever seen is up to this except in simple cases.
Take the New England Journal of Medicine. It's about $150 for an individual subscription and ranges from $1000 to $17,000 for institutions depending on the size. This is for a publication that doesn't pay authors, and in fact can make authors bend over backwards. No wonder all sorts of publication models are being explored.
Messing with the user agent string won't work, as googlebots use a well known IP range, and you can ID google like this. It's how we do, and of course it protects you from people _pretending_ to be googlebot but changing their UA string...
The other problem is that many legit sites, including my own, use 302 redirects a lot. They are used a great deal for maintaining backward compatability with old links when URL schemes change. They are also very useful in authentication and personalisation siutations
I don't get it. So they send these threatening letters to a few people - the so-called 'major abusers' or whatever, in the hope that the hundreds of thousands of 'home abusers' who just copy a few lines or, share their favourite algorithms with personal friends, are really going to change.
Frankly, if the GNU crowd want to stop copyright abuse, they'd better start outputting some source code actually worth protecting. I mean, have you _looked_ at sourceforge recently? It's all half finished projects and crappy PHP webapps. Who cares if we abuse the copyright on that stuff?
Basically, the new GNU license abuse is here to stay, and the Free Software industry better get used to it. Instead, they were lost in their own self-congratulatory world, talking up a few chart-toppers like Apache and Tomcat, while all around them hundrends of little start-up software companies were ripping the source code from apps and embedding in commercial stuff.
Basically, the FSF needs to ditch their bully boy tactics, and deal with the fact that copy-and-paste theft of neat bits of GNU code are here to stay. Sure, go imprison some college kids, if you want to stoop that low. But I for one will still be using peer to peer networks and the next generation of anonymous file sharing networks to obtain Free Software source code, and compile into my applications without telling anyone.
Yes, they are. Torrent sites are also legal in Sweden, for what it's worth.
"May be it is because Microsoft donates money and its software to the schools at subsidized prices that not many students learn about Linux?"
Oh, yes, that'll be it. Who'd learn about foo for free when they can get bar for almost free? I mean, the main reason Christianity is so popular in the US is because bibles are heavily subsidised.
Meanwhile, in reality land, it could be that Indian programmers would like maximise their chances of gainful employment, and so learn to use the most widespread OS? Gee, I wonder if that could be it.
Jon 'try the horse in front of the cart next time' Peterson.