I brought up the dates because we should quit wondering when things will happen and realize it already should have. it is a way of moving people down the path of free software, and realizing that our problems are social / software. Ray’s predictions are pessimistic, and that is an important point.

Give a man a fish, feed him for a day.
Teach a man to fish, feed him for a lifetime.
Make everything open source, feed the entire world forever.

I see techno-evolution happening on the commercial/proprietary front and on the opensource front. Proprietary software gets licensed out to various commercial entities, and on the evolution goes. Open Source evolution has less red tape and is more efficient however.

It is my opinion that the commercial side will eventually die, but not without a fight. Open Source grows by word of mouth, but Commercial Products utilize “Advertising” … I think the end result will be people will trust Commercial Advertising less and less as they see Open Source products perform better and better than commercial products.

One thing that Commercial Products have over Open Source is the 24 hour customer support — but even that is changing because it turns out people love helping each other and get a kick out of participating in online forums, chat rooms, etc. As chat bots get better, so will open source support improve. So really the Open Source movement will eventually solve it’s own problems. I believe in it :)

What makes me sick is there are over 50 “alpha” level projects in almost every software category that never get off the ground because of lack of commitment or interest. When no money is invested and no jobs are at stake, there is almost no commitment (besides the Linux fanatics & etc). Additionally if each of those 50 projects could unite their resources and passion, the project could actually take off and the whole world could benefit by actually getting the finished product!

What is the solution here? How can we unite 100 similar projects under one roof? It’s like trying to get fierce competitors to cooperate! How can we get more public contribution? How can we get more commitment? I have some ideas, but what are yours?
———-
From Keith:
Hello Chris Johnson;
I think Linux can take over before 10 years from now. But, if things don’t get better, it won’t take over in 10 years. Linux will only take over when it does a better job! This is a social question, not a technical one.

The success of Linux is a matter of getting this army moving more efficiently. Linux’s inherent advantage is the 10000 programs it has. Well, imagine when every one of those kicks ass? When will that be? I don’t know but, I do spend a chapter on the problems getting in the way, and the biggest problem is the bad programming languages.

There can be advertising with free software. You see lawyers advertise on TV, and many of their writings become freely available in law libraries. Free software never means that that the programmer didn’t get paid to write it. Linus has been getting paychecks to work on Linux for years.

You can have support with Linux. IBM, HP, Red Hat, and many other companies will provide you with support. Wave money in front of their face, and they will give you any SLA you could want.

I agree that there is a lot of duplicate code and dead projects in Linux. Writing more code in better programming languages will help. Using C instead of Lisp was the biggest mistake in the history of computing.

Evangelizing Python is the best way to get people to work better together. I spend two chapters leading to that conclusion. I also have another chapter on other problems with free software. The thing to realize is that we have more programmers than we need. We just need a few more to work together efficiently!

It comes down to having a good intuition regarding human nature with regards to the inherent value and limitations of open and closed systems. I quite enjoyed, although it is outdated now, Eric Raymond’s ‘The Cathedral and the Bazaar’ for its portrayal of the struggle between bottom-down and top-up design. Some projects are better with one, some the other, most with both. I agree with Ray in that they will continue to co-exist. I am optimistic in the fact that if open-source truly were better for everything then it would prevail. Or, that there is not some ‘fascist’ element to our societal power structures preventing the ‘best’ method to prevail in any extreme way. Of course there could be improvements, and those that hold the power (PhD professionals, closed-source companies) sometimes do prevent efficiency for the sake of maintaining their power, but ultimately, the process is quite democratic, and the best method will prevail. The best method, in my opinion, is coexistence. And ultimately, the two benefit from the other. I see a co-evolution taking place between open and closed.

In terms of making ANNs without having better knowledge of ‘how the brain does it’ we will have to agree to disagree on this. ANNs simulate some kind of intelligence well, but not others. I am confident in the fact that there are at least some underlying principles, ‘blueprints’ for ‘intelligence’ and ‘conscousness’, and that through reverse engineering the brain via MRI information we will discover them. The debate between ‘connectionist AI’ and ‘classic AI’ is still a debate for a reason. Like open and closed software systems, both methods have value.

I will pick up your book. Thanks for the discussion.

-Miley
———-
From Keith:
Hello Miley,

I have two disagreements with Ray’s sentence you quote above:

1. This isn’t about open source forms of information, only software. It is the “free software movement”.
2. Ray is saying that because proprietary software exists, that it SHOULD exist or that it is a good thing. I would reply again that starvation exists, but this isn’t a good thing. My book explains why free software is better for a free market than proprietary software, and necessary for us to solve the remaining big problems in computing, but people have to use free software. Ultimately, proprietary software and starvation will mostly disappear, but it involves humans making choices.

Not all choices are equally good. And taking away freedoms from software never makes the software better. Co-existence is not the best method, and it is slowing progress. Why haven’t we solved computer vision? Because there is no de facto codebase. Those people don’t realize that working together is important, yet. So while we can have this philosophical / moral discussion, consider the state of the software industry today.

I believe that anything can be simulated via ANNs. I just also understand how it would be very inefficient in some scenarios. It is like needing to add two numbers, and like inventing a programming language to do it in, first. I am quite sure there is nothing new we would learn from more MRIs of neural networks. All that is doing is taking more detailed picture of the biological machinery. It is like saying that taking pictures of your Intel processor allow you to make better software. We have the same problem with ANNs as we have for computer vision: where is the de facto codebase containing this code?

Thanks for the response.

There are a number of important differences between the Wikipedia project and the AI project. Wikipedia is a pooling of resources, where as with AI you are creating something for which there is very little precedent. AI projects have been unsuccessful for as long as they have because we don’t understand the underlying principles that govern intelligence. We don’t understand the brain. And consciousness, we really don’t understand. I see more analogies between the Linus project and the AI project in that regard.

That being said, a cruicial component to our meeting goals with regards to AI depends on our understanding the brain. And that is something that depends on hardware that we don’t in fact have, which is more sophisticated MRI technology. Faster larger processors that are capable of more so that we can scan the brain in greater detail in hopes of understanding the principles that guide intelligence. This very cruicial component of AI progress has nothing to do with software engineering.

Ray has never to my knowledge spoken out against open source software, and would be very cautious in considering that it is not factored into timelines he makes. He seems quite aware of the power of crowd sourced projects like Wikipedia, as they show up in his speeches.

So I guess I am curious as to why you think he does not factor in the power of open-sourced software into the timelines he makes?

All the best,

Miley
——-
From Keith: I have since gotten a reply from Ray and posted a reply back up above, and he and I talk about some of these issues, but I wanted to reply to your points as well.

I agree there are differences between Wikipedia and AI. Wikipedia is mostly data and very little code. I use the Wikipedia analogy because it is in the top 10 most popular websites, and well demonstrates the potential for mass collaboration. My first chapter is on Wikipedia, and my second is on Linux, which is a more thorough discussion. Note, there are people doing language AI research using Wikipedia, so the analogy is better than it first appears ;-)

I don’t agree we need more sophisticated MRI technology. We don’t need more detailed pictures of the biological hardware, we need a software re-implementation. There are many neural network implementations that are just hundreds of lines of code and have been kicking around for decades. When you digitize the algorithms of our analog system of computation, you can shrink things down a ton.

I don’t see Ray speaking out about free software very much. And in his letter above, he confirms that he sees it as equivalent to proprietary software. I know he doesn’t take the power of FOSS into his timelines, because if he did, he’d realize that we already have everything we need and we have for decades. He wouldn’t have timelines.

Regards,

-Keith

Suppose you had:

1) robot gardener & chef (= free food)
2) recycling robot (= free materials)
3) Efficient solar panels (= free energy)
4) open source hardware paired with a 3D printer and a manufacturing robot (= free physical products)
5) Open source software (= free software / AI)
6) Wireless (self replicating & self repairing) mesh network (= free internet & communication)
7) Flying Solar Cars (= free flights & transportation)
8) Robot scientists, doctors & nurses (= free health care)
(etc)

Some of these ideas are already available, or partially available. It doesn’t matter if these ideas will be fully available in 100 years or 200 years — It is plain to see we are headed in this direction.

My question is, when is the tipping point? When will money stop controlling the masses? When will people program cool software because they want to, not because they are getting paid to do it.

If I didn’t have to work for money, I would get into open source AI development to speed up humanity’s ability to solve all the world’s problems. Or I would get into genetics and stem cell research and develop immortality… but sadly I have to work for money!
———
From Keith: Hi Chris;

Not all FOSS has quality issues. There are many examples of free software that are higher quality than their proprietary counterparts. I spend some pages detailing how the LInux kernel is better than the Windows kernel.

I agree that when energy, computing and software are free, etc. we may move into a post-capitalist era where money will no longer matter. But this isn’t a binary situation.

Furthermore, free software is the most important next step, and it succeeds just fine in a free market. I spend a number of pages on this topic. It turns programming onto a service business like lawyers and accountants.

Maybe you can’t get paid to write free software, but millions of other people are contributing to it today, and a fair percentage are being paid. The problem isn’t that we don’t have enough programmers, but that they aren’t working together.

We need free software before we go into our post-capitalist society. Put another way, the fact that free software might succeed in a post-capitalist society doesn’t get us there so we shouldn’t consider it a solution.

It is a good question when the tipping point happens. But consider that the free software movement grows one person at a time. It is why I wrote this open letter. Some of the important and influential people can get others on board faster.

I wrote down a list of Linux world domination challenges here: http://keithcu.com/wordpress/?p=272

I agree with Simon. Ray’s views regarding what technology will accomplish in terms of intellectual tasks is considered by many in the software engineering field to be ‘optimistic’. Jaron Lanier is only one example. To say something even more ‘optimistic’, such as that these intelligent software platforms could already exist if only we were more willing to adopt a Linux attitude towards things, is immensely short-sighted. Its one thing to be in favor of more Linux like architectures in terms of software development, its a whole other thing to argue that very difficult computational tasks requiring very fast processors and ingenious engineering, could already be here if only we adopt that method.

———
From Keith: Science is not about what “many people” think. In areas of science not explored, there is no such thing as a consensus. Furthermore, I don’t know who is in this set of futurists you’ve polled for their pessimism, but it is much smaller than the set of computer programmers. My book starts with a quote by John McCarthy, the creator of Lisp:

“Some people think much faster computers are required for Artificial Intelligence, as well as new ideas. My own opinion is that the computers of 30 years ago were fast enough if only we knew how to program them.” (2004)

I believe there are many software programmers who don’t think the problem is a lack of hardware. Every programmer has seen magical improvements in hardware in their career! Their explanation for why we don’t have the technology yet is just that it takes more work on the chalkboard to have technical breakthroughs. That theory is also wrong, but for a totally different reason.

You can say I’m short-sighted, so I’ll leave you with a question: Pretend you aren’t aware of Wikipedia’s existence. How long do you think it would take for a group of volunteers to create the world’s biggest encyclopedia? Did you say 2.5 years? I realize that Wikipedia is different from Linux and AI, but the analogy is useful and I believe fully valid. It isn’t just Wikipedia and Linux that are successful examples of free software. There are many examples — but not in the AI / futurism world yet. This is the next step.

BTW, telling Ray Kurzweil that his predictions are “way off” by referring to your own book on free software is maybe a bit of a stretch don’t you think? Ray Kurzweil worked on software and AI for a big part of his life, just saying.

——
From Keith: Free software doesn’t preclude programmers being paid to write it. My book spends a lot of time on that topic. Linus Torvalds gets a paycheck. I added several more sentences to help clarify. There is no technical benefit from a lack of freedom. Free software isn’t better than proprietary software everywhere, but that is because it doesn’t have enough people working on it yet. That is a social problem.