Patents are a net cost to society, and software and business patents are teeing up to explode those costs. A large tax on the real economy this way comes, with the spoils largely to lawyers and financiers who produce nothing of real value.
But the recollections of the Marimba founder, and my memories of how people around me reacted to apt, both confirm my point that the technology seemed very innovative to most people at the time.
This had definitely been implemented and in regular use before 1995.
I am very familiar with reimaging machines from a central server, booting machines off of remote images, and roving profiles. All were in use before this technology. None fits the claims of the patent in any way. The key point of the patent being that it lets you get incremental updates to software whenever you want. The incremental bits avoids having to copy the whole thing, which is important when your connections aren't that good.
Incidentally rsync was publicly announced in 1996 so I very strongly doubt that they were using it in 1995.
That said, the patent filing is accurate in claiming that Java was not immediately used for menus. When we look back at things that far back it is easy to confuse things we saw and did a couple of years later with what was happening at that point of time. However it didn't take long to get there. Domino 4.5 (released in December of that year, under development when the patent was filed) definitely used Java applets for menus. Lotus 4.0 released in January of that year, did a lot of stuff, but don't know if they had interactive menus. Still they clearly had the idea, even if they hadn't implemented it yet. And lots of people were doing it by the end of 1997.
15 years is far too long a life for a software patent.
But I don't accept this premise. Even if automatically updating software over the network was prescient and innovative in 1995, that doesn't mean it would be prescient and innovative in 2000, or 2005, or 2010, even if this patent had not come into existence. I strongly believe that this kind of technique was inevitable as networks became ubiquitous, and getting the government involved in granting a monopoly on the technique is iniquitous and harmful to innovation.
In the first paragraph I open by pointing out that I don't like the existence of these patents. In the last paragraph I point out that the fact that the particular patent I selected is bad is an indictment of the patent system, and not the particular patent. Both statements run directly counter to any assumption that I think that government protected monopolies on once innovative ideas are good things.
In between I explain why, according to the rules for granting patents, this particular patent is relatively good. Note the critical qualification. I didn't make up those rules, nor do I agree with them. But those rules matter because they are the rules that this will be argued by in the courts.
And finally you aren't the first in this discussion to point out that this innovation was inevitable in the long run. To repeat what I said earlier on this point, the fact that I can point to two independent reinventions between the time this patent was filed and issued serves as strong evidence for that inevitability. However the argument that a given innovation was inevitable over time is irrelevant to any discussion of patents in the courts.
Take, for instance, patent #8, "Method and system for identifying and obtaining computer software from a remote computer" (see http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sec... for details). It covers a system for having lots of machines keep themselves up to date with a central repository. Obvious, right? We can all think of a ton of examples. For instance Debian's apt fits virtually all of the claims except the ones involving payment. (For some reason Debian doesn't worry about that much. :-) )
But look when it was filed. November 27, 1995. That's well before apt. In fact it was the same year that Debian first released dselect. I don't know if dselect was advanced enough back then match the patent. If so then that is prior art and I'd suggest that Salesforce's lawyers look carefully at it.
If it is not, then finding prior art may be a challenge. People today are used to subscription-based software distribution. But pick up Founders at Work and read the interview with Arthur van Hoff about Marimba, which was founded in 1996 (after the patent was filed) and was in the business of subscription-based software with automatic updates. He talks a bit about how new the idea was, and how it was difficult to get people to try it.
That tells me that, at the time and with the way people were used to working, this technology really was new and innovative. Now I really, really hope that someone can find prior art and smack Microsoft down. I don't want this kind of patent to threaten software I use.
But as far as patents go, this seems relatively good. (The fact that it is still bad is an argument against software patents in general, and not this particular patent.)