He thinks this could lead to courts handling "intelectuatual property crime" like shoplifting: First time no trial at all, second time a admonishment, third time a fine if you are not aware of the proper denial strategies.
Sure this argument might not be of interest if you life in a place where shoplifting is considered a serious crime ...
08:38 |
permanent link |
mail this
Sunday, 09. February 2003
Automated Antitrust Violations
I found out last week that CSS should be called "cartel enfocement mechanism" instead of "anti-piracy measure".
John T. Mitchell now has an article on
"Automated Antitrust Violations" basically working on the same line of thought.
[link credit: blat]
18:05 |
permanent link |
mail this
If we accept 'compound objects' of program and data how to handle that? IS is a program or is it data? How to decide? By percentage of bytes?
Are there Programs not created by man? Does creating a program really just can be writing some descritions in a "Programming language"? Can't a program also be painted?
Is recording a macro an act of programming? If I insist that it isn't - "you create a macro by clicking on buttons in your word processor, you donīt program it yourself - you simply tell the computer what kind of macro it shall program" (Olaf). I think this is a to limited view. Many big software systems are created nowadays with help of CASE tools which are not that different from MS Office's macro wizards.
If I create a C Programm - don't I just instruct the Preprocessor (or codegenerator) to write the program for me?
Isn't most 'logic programming' (right name? Traffic lights etc.) done by 'drawing' the logic?
It might be hard to decide if data structures are 'program' or 'raw data'.
I still insists Postscript files written by freehand/xfig/whatever are manmade programs. They express the authors intent to make a computer act in a certain way.
Perhaps creators of the law intended programs to be the stuff you don't change further and data the digital documends you further edit. This might fit the worldview of a 1985 lawyer.
Think about it again:
Perhaps when they created the law they didn't mean computer programs but
"computer data of any value." In fact this would make sense. Remind you: in the
80ies the only commercial available digital asserts where databases and programs.
Databases where not protected and later protected by special rules. Why did
programs need special rules? Because they are so easy to copy. Doesn't this apply
for all digital "stuff"? So perhaps everything is a program in legal sense.
Code never meant to be executed: Donald Knuth said once. "Remember I have only prroven that this programs are correct - I have never actually run them."
Could we reduce programs to binarys? No, see scripting languages, (Java|Python|Lisp|whatever) Bytecode, Microcode, CPU emulation etc.
Does a program need to be written in a programming language? This would not help us with code like /* this */
A classical 'binary' is not directly executable by the processor. The binary loader has to do a considerable amount of work to get it in a runable state. Think of dynamic linking, relocating, initialisation of segments etc.
Lambda the Ultimate reminds us that this discussion comes back to fundamental problems of computer science. See also me being cursed with the lack of a german term for 'computer science' - we use 'Informatik' instead but using a clumsy question I get an interesting answer regarding entropy as a clue to distinguish program and data.
Olaf: The work must be fixed in a tangible medium = This is, within our discussion, no problem. Nice talk I heard last summer on a "hacker conference". The speaker tried to keep data permanently "in flight" on the internet, e.g. by encapsulating it in a ICMP echo request (ping packet). Fascinating line of thought.
Olaf: ((4)) And, of course: If it shall be protected as a 'computer program' it must be a computer program. Doesn't the german law also state: "Computer Programs are programs of any kind"?
Perhaps when they created the law they didn't mean computer programs but "computer data of any value." In fact this would make sense. Remind you: in the 80ies the only commercial available digital asserts where databases and programs. Databases where not protected and later protected by special rules. Why did programs need special rules? Because they are so easy to copy. Doesn't this apply for all digital "stuff"? So perhaps everything is a program in legal sense. No "fair use" sorry.
We might not like that outcome but perhaps we should change it not by tinkering whith the definition of 'program' but in anoter way. Which Way?
11:53 |
permanent link |
mail this
I'm trying to understand how to use Creative Commons Metadata to encode non creative commons licenses. While checking the licenses at http://www.ifross.de/ifross_html/lizenzcenter.html I came to the conclusion that we will never ba able to express all this variations in a machine readable form. So I suggest that there should be a way to express at least a subset of the rights provided by the license.
The most usable probably would be "allows at least unmodified digital reproduction".
Many licenses also require that the "Disclaimer" is reproduced unmodified". Would be nice to be able to express this too. But reproduction would by its definition reproduce the disclaimer, too.
So for a BSD Handbook Licence I could use something like:
If we would be able to encode some of the popular licenses and could convince some of the producers of this works to add a simple "http://XXX/nameoftheirlicense.rdf" type="application/xml+rdf" />" to their HTML generating Tools we could significantly incerase the number of resources with machine-readable.license on the Web.
After poting my last Item I have studied the the http://creativecommons.cc site further and found out that they provide you with a some HTML-comment-encoded-RDF wenn you request the data HTML Data for the creative commons license button.
OK that is dead simple. (But still ugly) And further down in the document they imply that the Dublin Core elements are not mandantory - also they are not used in their own RDF.
To get this going it has to be dead simple to implement for content providers and convenient to implement for tool writers. First get rid of most of your fields or make them at least non-mandantory. You created a suite for encoding nearly all information related to IP for a certain document. Create a way to just encode a licence, without messing with variables for Title, Author etc. - so users just can drop some lines into their HTML without messing with template variables and the like.
KISS! Dublin Core Data does not follow the keep it simple style.
Get away from this HTML comment stuff. It is really, really hard to parse if there a dozends of them in a large HTML document. Regular Expression engines can get very slow at this and faster alternatives are a pain to code for. Besides being hard and slow to parse this commet thing can create a slew of other problems for tool writers. E.g. what to do if you see multiple licenses in a single document?
To allow inline definition of a license use HTML metatags or something like this.
The prefered way should be a seperate license file. Usually vou want to put several files under the same license. You also could link to the creative commons site.
Last night I had a long talk with the nice people at ifross. Seemes the IP-law reform this year will bring DCMA like provisions to germany. Some fun things from the current proposal of the "german DoJ":
If some technology keeps you from fair use, you are not allowed to circumvent it. But you have a right against the publisher that he provides the tools to you which are necessary to exercise your fair use rights.
Scholarship is protected by the german constitution which means this law can't keep university researchers form tinkering with circumvention devices.
Copying for research and teaching purposes is generally allowed.
