Summary: Jakob Nielsen's site has less text than some others, but he can still overwhelm the user with emphasized words. For each page users view, Nielsen delivers a different set of <strong> phrases to provide a distinctive-looking page.
We've known since 1997 that Nielsen's Alertbox site follows a roughly 10% ratio for random bold bits of text in the middle of a sentence. Simply stated, he tends to highlight large parts of the page, so much so that any real emphasis is lost.
The question is whether the same few phrases would always dominate, or whether the site would highlight different phrases in different pages. Studies now demonstrate that:
There is only 12% correlation between what is in <strong> tags and what is important.
Looking at smaller parts of a page, we see that the bold phrases tend to cluster together, with up to four of them in a single sentence.
82% of readers get so tired by the constant overemphasis that they do not read to the end of the page and go somewhere else.
I think really this is the fault of your browser - if the site just gives a
element containing some text then the browser damn well ought to format it so it is readable.
If Mr Nielsen has done something strange to deliberately screw up the formatting (as so many web sites sadly do) then you have a case. Otherwise, your complaint should really be addressed at the browsers. I completely agree, long thin lines of text are not very readable and the browser ought to line-wrap to 70 columns or so.
This is such an opportunity to post a good example of 'odd data' found on the Internet together with suitable jokes about 'back doors'. What's wrong with Slashdot these days?
The root of the problem is that any system relying on keeping your social security number secret is broken. An SSN is an identifier for a person, it is like a name. You don't keep your name secret (Wizard of Earthsea aside) so why should the number be different?
Not that you'd necessarily want people to be able to find out and disclose your number whenever they felt like it - there are still privacy considerations even with 'useless' information - but if disclosing the number exposes you to fraud then the fault is with the systems that rely on SSN to authenticate (rather than identify) an individual.
Every cheque you write has your bank account number on it. Disclosing the number doesn't automatically expose you fraud (unless you also supply headed notepaper and do other stupid things). If the banks can do it, why not social security?
Maybe all drink sizes should be abolished and we should order quantitatively - '200ml of coffee please'. This happens for most other goods and for beer.
If you search on Freshmeat you can find various 'distributions' for Mosix that mean you need only stick in a CD to have the machine boot up and add itself to your cluster. At least, that's the claim.
No, the 'novelty in technical contribution' criterion is meaningless in practice. Take a look at software patents already granted by the European Patent Office. These have been granted despite the fact that the current patent laws explicitly exclude programs for computers! The EPO has developed a spurious doctrine where a program may be counted as 'a computer program not as such', and thus patentable. Anyway, the current argument is about whether to amend the law so that the patents already granted by the EPO become enforceable. Given the EPO's past history of interpreting any rule as widely as possible, to allow as much as possible to be patentable, I don't have much faith in 'real technical contributions'.
What's needed is a simple, obvious rule that even the patent office can interpret correctly. The current European exclusion on 'programs for computers' is such a rule; the EPO has nonetheless managed to get around it by classifying some computer programs as not being 'computer programs as such' based on whether they have a 'technical contribution', but no court would accept such an argument, and it's courts that decide infringement. Rather than changing the law to match the patent office's warped interpretation, we should take the shocking step of asking the patent office to follow the law as written.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
>I don't think that computer programs are patentable (at present) in the >EU. In answer to your question,
It was more of a rhetorical question:-).
>no I don't think that courts will agree with the position taken by the EPO.
I was discussing this on a mailing list <http://liberte.aful.org/mailman/listinfo/pa tents> and one person responded:
>Firstly, whatever theory may say, the national courts have been very >happy to let the EPO Boards of Appeal make the running. The reason is >quite simple: few judges in the higher courts understand patents and >none understand software. In England, for example, there is no-one in >the House of Lords and the only Court of Appeal judge with patent >experience is Lord Aldous, aka Willie Aldous. In the High Court, Hugh >Laddie is a chemist; I don't know anything about Jacob, Pumphrey or >Neuburger, but none of them have afaik shown interest in the software >problem; at least, they haven't written on the subject. In Germany the >BGH is made up of pure lawyers who in my experience hate and fear >technical stuff. > >This is reflected in the way the House of Lords has talked up the >Boards of Appeal, referring to them as "judges in all but name". In >other words, the highest court in England sees the Boards of Appeal as >a court. In practice too, the English - and German - courts with the >exception of the occasionally perverse and consistently overruled 17th >Senate of the Federal Patent Court have sought to harmonise with BoA >decisions. The only disagreements I'm aware of in England arise from >older precedent, which the Patent Office feels bound by, in particular >Gale's Application.
So he seems to think that the national courts, the Lords in particular, will agree with the EPO's ruling because that's what they always do. I suggested that while the national courts are happy with the EPO's decisions made _within_ its remit (the EPC), they may not be so accommodating if the EPO seeks to extend its powers outside those granted by the Convention. But since there have been no cases so far it's all just speculation.
I wanted to point you to the archives of the mailing list but the recent messages have not yet appeared on the web page. I haven't mentioned you by name or email address but if you want I could cc you on messages I send, then you can join in the discussion. But you probably
Is it just me, or does this review amount to little more than "Ximian Desktop reorganised my menus nicely for me". It seems hardly worth posting on Slashdot at any rate.
One word: exams. If you're still writing individual letters separately by the time you sit written exams, you'll write at about half the speed of someone with good joined-up handwriting. In essay subjects it really helps to churn out long answers as fast as possible, and even in subjects with short answers it doesn't hurt.
Wiki-fying the jargon file would be cool - though there does need to be a real, live maintainer or committee of maintainers to weed out some of the junk, check references and produce an 'official' version (remember, the Jargon File is supposed to be a work of reference).
Myself I still think that redirecting to ESR's version would be better, no matter how much of an arrogant so-and-so he may be, deliberately breaking web links is going too far. Still a redirect to the new, ESR-free version would be better than nothing.
How the hell could the code 'be changed overnight'? Do you think that a secretive team is going to remove all copies of linux-2.4.20.tar.gz from ftp servers and replace them with doctored versions that lack the SCO code?
More likely, IMHO, is trying to migrate back to a patched earlier version of the kernel such as the 2.0 series, before the allegedly SCO-originating code was added.
ESR took over the Jargon File in the early 1990s. Before then it had lain unchanged for several years, so if you want an unsullied pre-ESR version you'd find it is mostly TOPS-20 and ITS and other systems which were long obsolete a decade ago.
I say let's just use ESR's version and apply due scepticism when reading it. Although it would help if the Jargon File, like a good dictionary, gave examples of usage with dates and authors. Then it would be easier to spot those that are ESR's own invention.
(If you do want to fork it, a coup would be to persuade the chap who owns tuxedo.org - who fell out with ESR a while ago and deliberately broke all links to ESR's site - to redirect ~esr/jargon/ to your new site.)
But 'faith in the code' is very often misplaced faith in your own abilities. Some programmers are true experts and can do as good a job as a decent library, but most cannot. It's a mistake to assume that if you wrote some code specially for a given task it will be better than a more general library. This is certainly true for things like the standard library (in whatever language), zlib, XML parsing, and other libraries that get a lot of testing. But even further down the suckiness scale you may find that a slightly buggy and not-too-well-documented library is nonetheless better than what you could write yourself in the time available.
The instinct to throw things away and start again is very strong. But you have to resist it! At the very least you should investigate what libraries are available for a task and play with one or two of them before making an _informed_ decision to roll your own. Most times I've seen wheels reinvented it has been through ignorance or through an (almost always mistaken) belief that the programmer knows better than the library developer.
It's about equalizing wage levels (slowly, but it is happening). For example textiles moved out of India long ago, because Indian workers are now paid more than they were. When it reaches the point that demand for workers in poorer countries starts to outstrip the available labour pool, wages there will start to rise too.
Yes, ZipFolders and the like do have side effects because they need to decompress to a temporary location. On RISC OS however (and I'm sure on some other systems too) you can run stuff directly from the archive, with no temporary files. Things are decompressed as needed. The archive appears exactly like a directory, that is you'd be able to open 'package.zip/README', edit it and save it. It's really neat and it's only a shame that a similar system hasn't been implemented for the free Unix-like OSes (which seem too stuck with the idea of 'mounting' a device onto a mount point somewhere, even for loopback).
Not lazy people, just people who don't know how to use the CLI or find it not to their taste. I use zip and unzip on Windows because I'm too lazy to install Winzip and wade through the graphical UI and nag screens to find the function I want. (The Winzip interface has always seemed rather perverse and too complex for the common case - creating or unzipping an archive to/from a single directory.)
A decent, well-thought out user interface for archivers is a different matter. I liked the system on RISC OS where navigation inside archives was just the same as regular filesystem access; in fact, you could even run applications straight from archives. That was because of pluggable filesystems, but the point is the interface shown to the users. (There are similar things for Windows - 'zip folders' or whatever - but somehow they seem clunkier. Maybe it's just bias.)
The Info-Zip 'zip' and 'unzip' programs are available for Windows, as 32-bit executables that handle long filenames perfectly. Why anyone would inflict WinZip on themselves is beyond me.
I'm confused. Does this mean you stick the CD into your computer, it makes a huge RAM disk, copies in the source code, compiles it all, and two weeks later you have a system ready to use right away?
Nonsense. Why should we take it for granted that 12-year-olds require supervision to use the Internet? They go to public libraries. I'd draw a distinction between deliberate acts like looking at 'bad' information (eg, pictures) and accidental acts like accidentally disclosing your location, getting your PC cracked, installing spyware not knowing what it did, and so on. The former probably can't be solved with technology, short of pressing the off switch, but the latter can and should.
The particular problem here is that of privacy. If the software broadcasts your name and location to just anyone then the software is not trustworthy. OTOH, if the software has some means of controlling the distribution of that information then it might be okay to run it. (I'm sure there are plenty of holes in the particular system I suggested about verifying the identity of other users before telling them your age, but anyway.)
Almost all interesting problems on the Internet are a mixture of technical and social. Take passwords for example. They can be socially engineered out of people, or read from sticky notes stuck to monitors. But still you should do all you can on the technical side to make them as secure as possible (eg, if they are random enough while still being easily memorized, perhaps people won't need to write them down).
Slashdot Mirror
Summary: Jakob Nielsen's site has less text than some others, but he can still overwhelm the user with emphasized words. For each page users view, Nielsen delivers a different set of <strong> phrases to provide a distinctive-looking page.
We've known since 1997 that Nielsen's Alertbox site follows a roughly 10% ratio for random bold bits of text in the middle of a sentence. Simply stated, he tends to highlight large parts of the page, so much so that any real emphasis is lost.
The question is whether the same few phrases would always dominate, or whether the site would highlight different phrases in different pages. Studies now demonstrate that:
element containing some text then the browser damn well ought to format it so it is readable.
If Mr Nielsen has done something strange to deliberately screw up the formatting (as so many web sites sadly do) then you have a case. Otherwise, your complaint should really be addressed at the browsers. I completely agree, long thin lines of text are not very readable and the browser ought to line-wrap to 70 columns or so.
1. Create a new Linux distribution that fills an important niche.
2. Put years of work and a sizable chunk of your life into it.
3. Don't Profit!
This is such an opportunity to post a good example of 'odd data' found on the Internet together with suitable jokes about 'back doors'. What's wrong with Slashdot these days?
The root of the problem is that any system relying on keeping your social security number secret is broken. An SSN is an identifier for a person, it is like a name. You don't keep your name secret (Wizard of Earthsea aside) so why should the number be different?
Not that you'd necessarily want people to be able to find out and disclose your number whenever they felt like it - there are still privacy considerations even with 'useless' information - but if disclosing the number exposes you to fraud then the fault is with the systems that rely on SSN to authenticate (rather than identify) an individual.
Every cheque you write has your bank account number on it. Disclosing the number doesn't automatically expose you fraud (unless you also supply headed notepaper and do other stupid things). If the banks can do it, why not social security?
Maybe all drink sizes should be abolished and we should order quantitatively - '200ml of coffee please'. This happens for most other goods and for beer.
If you search on Freshmeat you can find various 'distributions' for Mosix that mean you need only stick in a CD to have the machine boot up and add itself to your cluster. At least, that's the claim.
No, the 'novelty in technical contribution' criterion is meaningless in practice. Take a look at software patents already granted by the European Patent Office. These have been granted despite the fact that the current patent laws explicitly exclude programs for computers! The EPO has developed a spurious doctrine where a program may be counted as 'a computer program not as such', and thus patentable. Anyway, the current argument is about whether to amend the law so that the patents already granted by the EPO become enforceable. Given the EPO's past history of interpreting any rule as widely as possible, to allow as much as possible to be patentable, I don't have much faith in 'real technical contributions'.
:-).
What's needed is a simple, obvious rule that even the patent office can interpret correctly. The current European exclusion on 'programs for computers' is such a rule; the EPO has nonetheless managed to get around it by classifying some computer programs as not being 'computer programs as such' based on whether they have a 'technical contribution', but no court would accept such an argument, and it's courts that decide infringement. Rather than changing the law to match the patent office's warped interpretation, we should take the shocking step of asking the patent office to follow the law as written.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
>I don't think that computer programs are patentable (at present) in the
>EU. In answer to your question,
It was more of a rhetorical question
>no I don't think that courts will agree with the position taken by the EPO.
I was discussing this on a mailing list
<http://liberte.aful.org/mailman/listinfo/pa tents> and one person
responded:
>Firstly, whatever theory may say, the national courts have been very
>happy to let the EPO Boards of Appeal make the running. The reason is
>quite simple: few judges in the higher courts understand patents and
>none understand software. In England, for example, there is no-one in
>the House of Lords and the only Court of Appeal judge with patent
>experience is Lord Aldous, aka Willie Aldous. In the High Court, Hugh
>Laddie is a chemist; I don't know anything about Jacob, Pumphrey or
>Neuburger, but none of them have afaik shown interest in the software
>problem; at least, they haven't written on the subject. In Germany the
>BGH is made up of pure lawyers who in my experience hate and fear
>technical stuff.
>
>This is reflected in the way the House of Lords has talked up the
>Boards of Appeal, referring to them as "judges in all but name". In
>other words, the highest court in England sees the Boards of Appeal as
>a court. In practice too, the English - and German - courts with the
>exception of the occasionally perverse and consistently overruled 17th
>Senate of the Federal Patent Court have sought to harmonise with BoA
>decisions. The only disagreements I'm aware of in England arise from
>older precedent, which the Patent Office feels bound by, in particular
>Gale's Application.
So he seems to think that the national courts, the Lords in particular,
will agree with the EPO's ruling because that's what they always do. I
suggested that while the national courts are happy with the EPO's
decisions made _within_ its remit (the EPC), they may not be so
accommodating if the EPO seeks to extend its powers outside those
granted by the Convention. But since there have been no cases so far
it's all just speculation.
I wanted to point you to the archives of the mailing list but the recent
messages have not yet appeared on the web page. I haven't mentioned you
by name or email address but if you want I could cc you on messages I
send, then you can join in the discussion. But you probably
One word: exams. If you're still writing individual letters separately by the time you sit written exams, you'll write at about half the speed of someone with good joined-up handwriting. In essay subjects it really helps to churn out long answers as fast as possible, and even in subjects with short answers it doesn't hurt.
Wiki-fying the jargon file would be cool - though there does need to be a real, live maintainer or committee of maintainers to weed out some of the junk, check references and produce an 'official' version (remember, the Jargon File is supposed to be a work of reference).
Myself I still think that redirecting to ESR's version would be better, no matter how much of an arrogant so-and-so he may be, deliberately breaking web links is going too far. Still a redirect to the new, ESR-free version would be better than nothing.
If that attitude were universally adopted, everybody would be much better off. Sadly it isn't...
How the hell could the code 'be changed overnight'? Do you think that a secretive team is going to remove all copies of linux-2.4.20.tar.gz from ftp servers and replace them with doctored versions that lack the SCO code?
More likely, IMHO, is trying to migrate back to a patched earlier version of the kernel such as the 2.0 series, before the allegedly SCO-originating code was added.
ESR took over the Jargon File in the early 1990s. Before then it had lain unchanged for several years, so if you want an unsullied pre-ESR version you'd find it is mostly TOPS-20 and ITS and other systems which were long obsolete a decade ago.
I say let's just use ESR's version and apply due scepticism when reading it. Although it would help if the Jargon File, like a good dictionary, gave examples of usage with dates and authors. Then it would be easier to spot those that are ESR's own invention.
(If you do want to fork it, a coup would be to persuade the chap who owns tuxedo.org - who fell out with ESR a while ago and deliberately broke all links to ESR's site - to redirect ~esr/jargon/ to your new site.)
But 'faith in the code' is very often misplaced faith in your own abilities. Some programmers are true experts and can do as good a job as a decent library, but most cannot. It's a mistake to assume that if you wrote some code specially for a given task it will be better than a more general library. This is certainly true for things like the standard library (in whatever language), zlib, XML parsing, and other libraries that get a lot of testing. But even further down the suckiness scale you may find that a slightly buggy and not-too-well-documented library is nonetheless better than what you could write yourself in the time available.
The instinct to throw things away and start again is very strong. But you have to resist it! At the very least you should investigate what libraries are available for a task and play with one or two of them before making an _informed_ decision to roll your own. Most times I've seen wheels reinvented it has been through ignorance or through an (almost always mistaken) belief that the programmer knows better than the library developer.
It's about equalizing wage levels (slowly, but it is happening). For example textiles moved out of India long ago, because Indian workers are now paid more than they were. When it reaches the point that demand for workers in poorer countries starts to outstrip the available labour pool, wages there will start to rise too.
Yes, ZipFolders and the like do have side effects because they need to decompress to a temporary location. On RISC OS however (and I'm sure on some other systems too) you can run stuff directly from the archive, with no temporary files. Things are decompressed as needed. The archive appears exactly like a directory, that is you'd be able to open 'package.zip/README', edit it and save it. It's really neat and it's only a shame that a similar system hasn't been implemented for the free Unix-like OSes (which seem too stuck with the idea of 'mounting' a device onto a mount point somewhere, even for loopback).
Not lazy people, just people who don't know how to use the CLI or find it not to their taste. I use zip and unzip on Windows because I'm too lazy to install Winzip and wade through the graphical UI and nag screens to find the function I want. (The Winzip interface has always seemed rather perverse and too complex for the common case - creating or unzipping an archive to/from a single directory.)
A decent, well-thought out user interface for archivers is a different matter. I liked the system on RISC OS where navigation inside archives was just the same as regular filesystem access; in fact, you could even run applications straight from archives. That was because of pluggable filesystems, but the point is the interface shown to the users. (There are similar things for Windows - 'zip folders' or whatever - but somehow they seem clunkier. Maybe it's just bias.)
The Info-Zip 'zip' and 'unzip' programs are available for Windows, as 32-bit executables that handle long filenames perfectly. Why anyone would inflict WinZip on themselves is beyond me.
6. Profit!!!
You mean Ximian deleted the 1.4 packages from their FTP site? Are they not mirrored anywhere?
I'm confused. Does this mean you stick the CD into your computer, it makes a huge RAM disk, copies in the source code, compiles it all, and two weeks later you have a system ready to use right away?
Nonsense. Why should we take it for granted that 12-year-olds require supervision to use the Internet? They go to public libraries. I'd draw a distinction between deliberate acts like looking at 'bad' information (eg, pictures) and accidental acts like accidentally disclosing your location, getting your PC cracked, installing spyware not knowing what it did, and so on. The former probably can't be solved with technology, short of pressing the off switch, but the latter can and should.
The particular problem here is that of privacy. If the software broadcasts your name and location to just anyone then the software is not trustworthy. OTOH, if the software has some means of controlling the distribution of that information then it might be okay to run it. (I'm sure there are plenty of holes in the particular system I suggested about verifying the identity of other users before telling them your age, but anyway.)
Almost all interesting problems on the Internet are a mixture of technical and social. Take passwords for example. They can be socially engineered out of people, or read from sticky notes stuck to monitors. But still you should do all you can on the technical side to make them as secure as possible (eg, if they are random enough while still being easily memorized, perhaps people won't need to write them down).