Ali's Journal

In the past, I wrote about copyright and openness. It was said that there should be a mechanism where open innovation is profitable for those who innovate so they are encouraged to keep innovating. The idea of copyright is to make sure innovators have incentives to innovate but I discussed the negative aspects of closeness.

Clearly, I always support openness but the prerequisite for openness is to make sure, it still encourages innovation. Until a robust mechanism replaces copyright mentality with an open mindset, I’ll take the advice of those who say I shouldn’t write everything here before the overall platform is complete. I personally think writing everything here is a good practice because this project is tight with the reputation system and other elements of the platform and if other people copy one element of the project, it’s not really very important because they cannot go on and developed the reputation system and other crucial elements of the project.

I’ll work a few more days on the basic ideas and then make a presentation of the project (phase one) and contact the Guardian.

UPDATE 1:

It’s been pointed out that generative AI might produce non-sensical outputs. If it’s fed with wrong, biased inputs it still generates an output based on them. The other approach can be based on an evolutionary mechanism where some strategies (or inputs) retain their development while competing strategies (or inputs) stall. As a result, non-sensical inputs are pushed aside by the evolutionary mechanism, but more coherent inputs will be much more influential in the output. So AI generates several competing initial outputs, and then they compete with each other and evolve.

“Coherent” is very unclear here. The evolutionary mechanism is unclear too. this project can be a social experimentation to guide researchers on how it might work in a social context. Currently, it’s unclear how generative AI produces outputs. This is one of the main concerns about AI. This project is the opposite. It doesn’t incorporate generative AI very much (at least in the initial phases), instead, it constructs several layers (similar to deep learning) but each layer consists of groups of people. The aim is to understand which evolutionary mechanism works best in this environment and if it can be used in generative AI.

(What I said about openness doesn’t hold for research and scientific topics – It will have very positive impacts if the scientific community, all around the world, is based on open/ democratic processes. When this approach proves to be more productive, even authoritarian systems will welcome this approach in certain areas).

Constructive debates are the ones where one side brings inconsistencies of the other side’s arguments to the foreground and tries to defeat the other side’s point of view. Endless debates, on the other hand, are the ones where each side insists on their sides consistently without any clear winner. In evolution, it’s critical that some strategies defeat other strategies. the same should be true in this platform. That’s comparable to a battle in court. The defence team for example might come up with a trap for the other side hoping that they might build on top of that and then reveal the inconsistency. Or they might incorporate other strategies. The aim is to win over the other side’s viewpoint. The defeated viewpoint is like a defeated troop they might reorganize and deploy a new strategy for a comeback.

The battle to evolve a point of view discussed here is inspired by court battles. Each side should look for inconsistency or potential attack points to the other side’s viewpoint in a systematic way. All people, every day, use algorithmic thinking. But they don’t know it. For example, they might say “If I can get out of home before 8, I’ll walk to work, else I’ll use public transport'” This is very basic algorithmic thinking. If they do it knowingly they can build on top of that and design more complicated algorithms. The same is true about debates. If people use systematic methods to reveal inconsistencies and shortcomings of the other side’s viewpoint, they can develop that. So the game mentioned yesterday is very similar to the conventional strategy games but instead of troops and things like that, gamers are working with arguments.

I will apply for the Global Talent visa. Before that, I should contact different people and organizations. This is a multidisciplinary project and involves many organizations. First I’ll contact the Guardian and discuss the idea mentioned yesterday for a test scheme on ways the community can evolve an article. This requires collaboration with AI specialists. So I’ll contact people in that field too. Mustafa Suleyman is of course among the people I’ll contact with. Then will send an email to an economics professor in Oxford and discuss the economic aspect of the project. Read Jean-Paul Carvalho’s profile and I think he is a suitable person to contact. Then it’s important to contact design and music colleges because the project will not be like chatGPT averaging numerous inputs. There are more elements to it than simply producing a summation text.

Meanwhile, I’ll work on the math/physics articles too. In the morning I’ll continue working on the articles and spend afternoons communicating with people in different areas and disciplines.

UPDATE 1:

It’s clear that collaboration with the Guardian/Observer should not imply that the project belongs to any specific party or political spectrum. The aim is to gradually expand the project so it includes more and more institutions.

Writing an article is similar to writing a code. A programmer initially writes a crude sketch of the overall program and then works on the details. A writer does the same. He/she draws a map of the article. For example, simply speaking, they decide there are two main arguments to the story. A, B, and C support argument 1 and D and E support argument 2. Then A, B, C, D and E are developed (for example A1 and A2 construct element A and so on). So in a sense, a writer draws a map of the article. Not all steps might be written down nevertheless the process of writing an article resembles the one described above.

In a conventional article, a writer tries to make sure the final output is complete and coherent. All elements of the story have been polished and worked out thoroughly. In contrast, in an interactive article, the writer intentionally includes incomplete elements and empty boxes. These elements specify the possible trajectory of the evolution of the article. So in some ways the writer, very roughly, designs the evolution path. The same holds for users who add inputs. They too add incomplete elements as a nudge for other users indicating in which direction the evolution should take place.

At first, only a few interactive articles should be written. This is an experiment to see how things work in this setting. The aim of these types of articles is to develop a type of consensus mechanism for ideas, and therefore should eventually reach some sort of conclusion. The next phases of the project will involve more practical approaches to tackle the problems discussed in the article.

UPDATE 2:

This should be a real game. It means some companies fund the project (as a type of advertisement) so that users get paid for influential contributions. So users compete with each other to push the article toward their intended evolution path.

UPDATE 3:

It’s unlikely that isolated inputs make a huge difference in the overall evolution path. The same is true in nature, a sequence of mutations and genes make an observable difference. So a minimal forum platform should be developed so that users can team up and discuss ideas before they insert their input. So this is a semi-text-based game and incorporates researchers from various disciplines (social sciences, biology, creative fields, etc).

UPDATE 4:

As I said it’s important that the project is politically independent. I think the organizations and people I mentioned before don’t contradict that aim. Anyway, I think the best option for me is to apply in the tech branch but I will first contact a music professor I knew some years ago and discuss the project with him. Then start the project and collaborate with different people and organizations before entering the UK. When the project demonstrates tangible progression then I can actually apply for the visa. But, like any other project, one of the first steps is to find sponsors, because apparently, I cannot enter for a while and I should travel to different countries for some months. If the project starts I can meet and collaborate with people in different countries. I’m leaving Turkey (but I may be back), but if the project starts I can collaborate with people living in UAE for as long as I stay there. So I think the first step for me is to present the project and contact my previous teacher as the first point of contact.

The other option is to complete the math article and apply for the research visa. I will do both concurrently.

UPDATE 5:

This project tries to change journalism for good and starts a new path in this area, so it’s important to stick to the initial plan which includes the Guardian.

Social media content can be extended differently (writing comments, sharing, retweeting, etc). I wrote something a while ago about seamless content. In such a platform, the evolution path is critical. Social media content shouldn’t be evolved randomly, instead, there should be a mechanism that is responsible for how things evolve. As it was said in the past, writing hundreds of comments for an article is not a good evolution strategy. But an article is mostly text, by generative AI, a text can be evolved into multidimensional content. Then users should come up with strategies to add something to the content in a way that is supported by the evolution mechanism. So each update is similar to a mutation but those mutations should be supported by the mechanism so that they evolve more and more. It means users shouldn’t randomly write a comment. That’s like a strategy game. Each update should be added with its evolution path in mind.