True, Alan Kay did not invent object-oriented programming and then looked at those ideas and made the internet.

In fact Alan Kay recognize this and has taken pains to point this out. He said he did not invent OOP, he merely discovered it. In his mind OOP was something already happening all over in different areas. He simply attached a name to it and described it.

I suppose it is like creating a taxonomy of animals. The mamals, reptiles and birds are already there. But there is still some value in somebody coming along and organizing animals into these groups and sticking labels on these groups.

I think what Kay tried to do with Smalltalk was to distill the essentials of these ideas and see how far you could take them. Early Smallalk e.g. was no merely a progamming language. It was a whole world onto itself. There was no OS in the sense we are used to it today. There was no applications.

It was a computer running only Smalltalk. Everything you may have thought about as an application, was simply a Smalltalk object. The Smalltalk IDE was not a window into just a little sandbox with your project. No, it was a window into the whole system. With it you could modify ANY part of the system. Including anything you views as an Application. That included even the IDE itself.

Everything was modifyable, and you could modify it live. If you wanted to change how your text editor or image editor worked, you could do that live.

All the objects making up a text editor, where available to you do compose another text editor or another type of program. You could probably just reuse the whole text editor object if you liked.

In effect he tried to recreate this live system that exist on the internet inside one computer. A place where any program could be modified and made to interact with anything else. And where all this could be done live and with immediate effect. In such a scenario encapuslation and message passing is essential as you have the ability to change the whole system.

I think this is Alan Kay's objection to how the Web developed. When he says internet is an OO system, he means the servers and the TCP/IP protocol. At that level. He is far more critical of the Web as such.

The Web is creating a lot of what Kay did with Smalltalk but in a far more limited fashion. We need special software, the web browsers. Web apps live inside this sandbox. The Alan Kay vision would have been more like the whole OS being just one big web browser. Like every object would exist at the same level. There would be no fundamental difference between web apps and local apps.

I think something like Plan 9, is more in line with the Kay vision. In Plan 9, apps would work in a sever client model. They would create mount points in the filesystem to expose their functionality. E.g. a Window system would create files representing the mouse, keyboard, windows etc. Clients would then read and write to those files to display themselves and catch mouse events.

But these mount points got creeated through a distributed protocol named 9P, which meant a Window server could be running on another computer and create mount points locally. Thus a local client could interact with the mouse, keyboard and window device files on your computer but it would create a display/representation on another. In this regard, you remove the artificial barrier between computer on the network. It is all just message passing, whether local or remote.

In todays OSs these worlds are kept very isolated. I think that is Alan Kays disappointement. Plan9 also hightlights that he merely indentified the pattern. I don't think the creators of Plan9 really ever read Alan Kays ideas of tried to replicate them. They just arrived at similar conclusions and use different terminology. They never labeled what they do as object-oriented, even if in the Alan Kay sense it really was.