Copyrights are going to be forgotton

Either copyright should apply equally to all works, or it should apply to none,

I disagree. We should be careful in determining which classes of works we should grant copyrights to, and which we should not, and what the maximum lengths of the terms for various classes of works should be.

This is because copyright is meant to be an incentive to create and publish works, but as minimally restrictive upon the public as possible in scope, and of minimal length so that the work enters the public domain as quickly as possible.

Ideally a copyright would be exactly the bare minimum that was required to get the author to create his work; this might not be what the author would want, but what the author could live with. Unfortunately, the Copyright Office lacks the large number of mind readers they'd need to do that.

But we can still get fairly close.

We know that copyright is merely an economic incentive. A copyright won't make an author famous or well respected or anything. It doesn't even guarantee wealth; like a lens it merely focuses some of the wealth that can be derived from a work, with the actual amount determined by the market. Thus a copyright on a flop is just as powerful as a copyright on a smash hit, but the flop has less money to concentrate to begin with. Copyright isn't even the only economic incentive; plenty of authors get paid for their labor without needing copyrights, and plenty of authors can sell specific copies of works without needing copyrights to protect them (e.g. anyone who could and would buy a painting from Picasso was not going to buy a cheap poster as a substitute; they wanted a specific copy, not any copy).

We know also that most works have no copyright-related economic value at all. Of the few that do, most are of modest value, and most of that value is concentrated in the first little while after publication in a given medium. For example, a movie has most of its ticket sales on its opening weekend. Each week thereafter, sales usually die down. Eventually it no longer sells enough tickets to remain in the first run theater. It goes to the second run theater and again, has brisk sales at first, and then winds down. It goes to Pay Per View, same thing. It is released on home video for rental and sales, and again the same thing. It goes to premium cable networks, then basic cable networks, then broadcast networks. Eventually it's lucky if someone wants to show it as the late night movie at 4 am.

All works go through this sooner or later. A daily newspaper is only good for birdcages or fishwrapping as soon as the next day; a basic math textbook can last for decades. Most works hang out somewhere in between. A work that has lasting popularity is as rare as a winning lottery ticket, and even it has a cycle to it (I have a copy of Star Wars, if I buy another copy, it'll be a while).

So since copyright incentivizes authors according to the money they can make, and since most of the money a work will ever make is made pretty quickly, with just how quickly varying based on what kind of work it is (book, sound recording, movie, newspaper, computer program, etc.) we can vary the term length based on the kind of work and thus provide almost as much incentive as we do now, but at a far lower cost to the public by reducing term lengths, so that the work is in the public domain sooner.

So first we need to require authors to apply for copyrights for their finished or published works, rather than just getting them automatically (I'm fine with modest protection to unpublished works in progress, so that the author need not fear someone running away with his manuscript, but even that should have some limits, and be no substitute for a real copyright on a published work). This way authors that are not incentivized by copyright at all will probably fail to apply, even though it should be quite easy, and the public can reap the rewards immediately. Authors who are incentivized by copyright will likely care enough to fill in the simple form and pay the token payment, which is just there to discourage abuse.

Let's say we grant a 2 year copyright term, renewable by the copyright holder x number of times, where x is what we are varying. A copyright holder can always not bother to renew at all , which is good, since it gets the work in the public domain sooner. Renewal will be like the initial registration: a very simple form and a token payment, just to prevent people abusing the system; even the mildest active interest in a work is fine, so long as we aren't rewarding total apathy.

Newspapers and other periodicals don't get renewals, and probably wind up with an overlong copyright anyway. But who buys 2 year old magazines from the publisher (as opposed to used) other than a doctor's office?

Computer software also 'ages' pretty rapidly. People like to run the newest software, and developers like to improve their software to convince people to buy it again. MSDOS is still copyrighted, but I doubt Microsoft cares about that in the least; no one buys it anymore, because no one wants it. So how about 2 renewal terms, giving software a maximum copyright of 6 years (provided the copyright holder renews bianually)? Yes, this would mean that you could freely make copies or non-GPLed forks of software from earlier in this decade, but so long as it would not discourage developers from making their new software, that's fine. And if they can't make something that can outsell free copies of their old work (e.g. Vista v. XP) then that's their problem. Now they've got even more of an encouragement to always outdo themselves.

Books, music, sound recordings, and movies all age pretty well; we could let them renew up to a maximum of, say, 9 renewals, or 20 years.

And some works are even more interesting. Until about 20 years ago, the US did not grant copyrights in buildings, but eventually we were required to because we entered into some astonishingly stupid treaties (which we'll need to get out of if we want to reduce terms below life+50, require registration and other formalities, etc.). Yet even though for the majority of our history buildings were uncopyrightable, buildings kept on getting built. Even really good buildings, like the Empire State Building, or Grand Central Terminal. And now that buildings are copyrighted, we haven't seen a surge in building that is attributable to copyright. Improvements in engineering, trends in architecture, the economically productive use to which the building will be put, etc. are all incentives, but copyright simply is not. No one is pirating buildings outside of the cookie cutter suburbs, and frankly, that hasn't stopped suburban development yet (which is too bad).

So we might as well just abolish copyright on architectural works. It incentivizes no one, and harms the public just by existing, so it is totally lacking in good features.

In short, copyright should be carefully tailored to works, or at least classes of works, rather than being equally applied to every damn thing. Agreed?