Seminar with Alan Kay on Object Oriented Programming (VPRI 0246)

OMG, what a bunch of baloney. Kay clearly knew nothing about either biology or computer science. No wonder that he appeals to people who have absolutely no knowledge about programming.

Decouple Algo step

Thanks for sharing this video 😊

与艾伦·凯（Alan Kay）讨论面向对象程序设计（VPRI 0246）

This is like seeing OOP God. Great respect for Alan Kay Sir.

Does anyone know what year this is?

Given his mention of ParcPlace Systems and Smalltalk/V Mac it's at least 1988, I'd say 1989?

this isn't c++ reeeee

Gold diamond ruby Platinum

Man Alan Kay is a great mind. But he's also so full of shit lol

Re. "subclassing class-class" - Maybe people could learn about this by looking at the Smalltalk/V manual he talked about, but I felt like the part that always gets left out of these presentations of Smalltalk's meta capabilities is how to modify the UI to go directly at what you've changed. In this case, you'd probably want to modify the browser, so that you could work with your new kind of classes. Though, I have heard about how people have come up with browser frameworks in Squeak, for example, that allow a high level of customization (doing I quick lookup, I see an old presentation on building a custom browser in Glamour).

So this can be summed up by a Teddy Pendergrass song: "You got, you got, you got what I NEED!" :D

This talk was in 1987

No audio. Please fix.

Creating future

Thia video's date is uncertain the uploader thinks it's from the mid-1980s. It was uploaded in November 2017 and it has only four thousand views. There are only four comments. This astounds me.


 A lot of the talk is about projects in trouble. In such projects one common thread is an unspoken wish to change underlying ideas of the system after the system is half built. The result  in SQL systems is usually table creep.


 At an elementary level I think he has got some things wrong but so have  some supposed relational people when they think the only value of object application coding technique is data type inheritance.


Both camps miss an important concept, the idea that a system needs to be coherent and consistent. By insisting on a separation of language from data for example relational language and host language EF Codd as early as 1970 recognized what was missing.


 he recognized that in effect base relations are the axioms of a system. the SQL industry goes wrong by assuming a system should be able to change axioms in mid-flight. If the object industry thinks applications can enforce system-wide axiom's then they are wrong too.

On the other hand Alan K makes the screwy claim that a relational system cannot tell you how old somebody is, only their birthday. The object business seems to think that relational versus object is a binary choice. The problem of storing axioms can't be solved relationally they want some kind of roll-your-own persistence.


 When an application that is already written depends on certain axioms for example the laws of two-dimensional geometry why should it be forced to be Rewritten because some other application needs different axioms?


 It all reminds me of some exchanges elsewhere about relational theory. One poster claimed that a relational tuple could change. This reminded me about the old joke of how many psychiatrists does it take to change a lightbulb. The answer is only one but the light bulb has to want to change.

Now, this is priceless. Thank you, Ohshima.

This is a master piece!!! Thanks Yoshiki for sharing it, very much appreciate it.

Thank you!