Alife News

The Artificial Life Community Newsletter

A word from the team

Welcome to the 6th issue of the ALife Newsletter. Summer is here in the North hemisphere and we are all being boiled alive, but hopefully this fresh newsletter will help cooling you down!

This edition has several paper reviews, a course review, and two workshop summaries! If you attended an ALife event recently, let us know.

The Newsletter is distributed by email, and archived on the International Society for Artificial Life's website.

You can subscribe here.

Contribute to any section of the next newsletter here! We are specially interested in Master and PhD students who want to talk about their own research.

Lana, Imy, Mitsuyoshi, Claus and Katt.

Review of the Agent Learning in Open-Endedness Workshop

By Kevin Frans

Open-endedness is the idea of a system that "endlessly creates increasingly interesting designs". That's a simple concept with big implications, especially for those of us interested in AI and the emergence of intelligence. Because even with powerful learning algorithms, intelligence is one of those tricky things that's hard and complicated to specify. But that's the beauty of open-ended systems: they can generate unexpected designs, designs that go beyond the intention of the designer. And so even if intelligence is hard to create, it might be possible to create a system where intelligence naturally emerges. This weekend, a bunch of us thinking about how open-endedness can boost AI gathered at the Agent Learning in Open-Endedness (ALOE) workshop at ICLR.

Many projects were presented, mainly centering around the question of "how can we generate training tasks that are rich, learnable, yet increasingly complex?". XLand aims to create an open-ended set of tasks by 1) defining a distribution of games, and 2) making them all multi-agent. The key lesson is that XLand tasks are so varied and dense that agents have no choice but to generalize. Sam Earle presented work on procedural content generation, showing that AI models could be used as level generators, resulting in a competitive game between AI players and AI designers. PAIRED, presented by Natasha Jaques, examines the game theory of this framework, and shows that a strong training curriculum can arise from a game between a designer, an antagonist, and a protagonist. The designer tries to propose tasks which the antagonist can solve, but the protagonist cannot, a metric which empirically corresponds to tasks just on the learning boundary. Finally, Julius Togelius argued that our classic framework of "agents interacting with an environment" is too limiting, and we should explore new ground.

For more detailed notes, check out the full blog post

Artificial Life Reading List

By Anya E Vostinar

This is a reading list of sci-fi books (and a few tv shows/movies) that explore humanistic alife topics for a college-level reading-intensive course. Questions include 'What makes an individual?', 'What makes us human?', and 'Are androids human?'

Artificial Life Reading List

Running a Life Simulation for 60 hours!

by Claus Aranha

Remember The Bibites, a life simulation that we introduced back in Edition 04?

A few weeks ago the simulator received an uppdate, so I decided to let a simulation run in my work computer for the week. It was very nice to check on how my bibites had evolved during my work breaks.

30 hours of bibites

the simulation at around 30 hours

Because there was no physical separation in the environment where the digital creatures lived, the population quickly converged. However, evolution continued, and it was very interesting to guess when a new mutation would take hold of the population (such as the ability to avoid other bibites), or be lost to genetic drift (such as the expression of colorful pheromones).

By the way, the author of the simulation recently released a similar video, where he follows 100 hours of simulation and prepares a simple phylogenetic tree of the species that came up. He also promised better tools to visualize the simulation, so look forward to that!

Course Review: Origins of Life (by Complexity Explorer - Santa Fe Institute)

By Imy Khan

For a couple of years now, I have sought to improve my understanding of the fields of origins of life and astrobiology, as well as refresh (and/or improve) my working knowledge of physics and chemistry as they relate to living systems. This desire was both a personal one (I think origins of life and astrobiology as research areas are just cool) as well as “professional”, given the overlap between those fields and Artificial Life.

Many readers may agree when I say that diving into and (attempting to) teach yourself new areas can be a daunting and overwhelming endeavour. When it comes to learning about new areas and disciplines (often outside our areas of expertise), knowing where to start or what the foundations of the field are, and then to source the “best” material to learn the new concepts and theory, is a task that perhaps leads many people to either give up their attempts to learn altogether (or, alternatively, creates gaps in their foundational knowledge that need to be filled in later on). Herein lies the value of structured courses, in my humble opinion, where these problems are solved by experts in the area of interest. So, I was quite excited when I found that Complexity Explorer (Santa Fe Institute) was running an Origins of Life online course which started in June 2022. It was even nicer knowing that the course was free of charge (which, to the best of my knowledge, is true of all Complexity Explorer courses).

I chose to write this short review of this course for the ALife community, as I suspect I may not be the only person who may be interested in better understanding OoL/astrobiology or some of the related disciplines. In fact, studying “origins of life” isn’t strictly correct. Rather, what we are doing (I think) is learning about concepts and principles from other disciplines - including the earth sciences, biology, chemistry, and physics - as they relate to the emergence of the properties of what we call “life”, and then to question, explore, and try to figure out what the minimal (biological, chemical and physical) conditions might be in order for “life” to emerge. Or, what “life” would look like if those properties or parameters were different. That second point might make the tie-in to ALife a little clearer…! (There may be more to it, but so far, these are the things that have me interested in OoL research.)

I have been apprehensive about committing to online courses in the past, particularly as so many currently exist that it’s sometimes difficult to separate the good from the…less-good. But since Complexity Explorer/SFI have a reputation for delivering high-quality courses, I decided to take the plunge and commit to the course for the entire duration (11 weeks). Despite the fact that I have no formal background and very limited exposure to anything related to OoL/astrobiology, I have found this course to be a very accessible entry point into this field of research. The course instructors (Prof. Chris Kempes (SFI), Prof. Sarah Maurer (CCSU), and instructor Maria Kalambokidis (UMN)) have done an excellent job of putting together a well-thought-out roadmap of content in both the weekly topics (each week’s subtopics are carefully crafted and structured to build on top of each other) as well as throughout the course itself. Lectures have, so far, been easily digestible, well-presented, coherent and succinct (the length of each lecture is anywhere between 8-25ish minutes each). The instructors have also been thoughtful enough to provide a comprehensive list of references and supplementary material related to each week’s topics, addressing the previously-described problem of not knowing where to start when it comes to finding information on new topics. Crucially, the course has no prerequisites, though they do mention that it may be useful to have some background knowledge of algebra, introductory chemistry and biology. Though I have little knowledge of chemistry, I have not been overwhelmed by the chemistry-heavy units so far, but rather, found them fascinating and insightful. Perhaps this is indicative of how thoughtful and meticulous the instructors have been in choosing how to collate and present the information.

In terms of time commitments, I have been able to watch all the lectures, read the material, and make study notes accordingly within 3-4 hours per week. This is a fair commitment, in my view and like with any course, I suspect you will get out what you put in. If you are interested in learning about (or improving your basic knowledge about) OoL concepts and theories, but don’t know where to start, then this course gets my recommendation. There are (optional) assessments and quizzes to test comprehension, with a certificate of completion if you meet a certain mark threshold, but the course can also be completed without submitting any of those. Nevertheless, if you can carve out some time (3-4hrs per week) to commit to it, I think it will reward you accordingly. At the time of writing, I am on week 3 of 11 (this week’s topic is “Chemical Commonalities” which, amongst other things, will discuss some of the common chemical properties across different living systems, DNA as information, and chemical configurations) and I fully intend on continuing this momentum for the entire course.

Overall, this course is playing its role as an introductory course brilliantly, giving me a stronger understanding and appreciation of some of the core concepts and principles of the field. As a result, I suspect this will become a jumping-off point for me to be able to confidently dive into OoL research in the future with a more informed perspective.

ALife Paper Review

"Evolutionary transition from a single RNA replicator to a multiple replicator network"

by Mituyoshi Yamazaki

This research is an attempt to realize a host-parasite coevolution model aiming at Open-Ended Evolution.

Research Content

The following two conditions are necessary for evolution:

How to build a simple model that meets these conditions?

Previous Research

In Spiegelmann's previous study, self-replicating RNA was evolved by feeding it raw material and an RNA-replicating enzyme to replicate it, and repeating the same procedure for the next generation. As a result, RNA converged on the shortest gene that has a trait with high replication efficiency.

Cause of convergence: - The selection pressure of RNA did not change because only RNA was passed in each procedure and the same type of RNA replication enzyme was added. - The information encoded by RNA was meaningless (the genetic information encoding the protein was included but not used for replication procedure) - The susceptibility to replication depended only on the physical properties of RNA and the replication enzyme.

Research by Professor Ichihashi

In this paper, Professor Ichihashi continues the previous work by introducing a cell-free protein synthesis, and implementing a model in which RNA makes an RNA-replicating enzyme and the replicating enzyme makes RNA. RNA mutations also affect replication enzymes, avoiding convergence by changing the selection pressure of the system.

Along with that change, they introduced a cellular structure. (In order to prevent the generated RNA-replicating enzyme from drifting somewhere, the cell structure pre-encloses RNA and the necessary nutrients. If it drifts, the tragedy of the commons arises).

As a result of conducting experiments on such a system, the introduction of a cell-free protein synthesis system was not enough, and it converged and stopped evolving.

At this point, the parasite had already developed, but it was suppressed to not inhibit the host's self-renewal. After removing the repression and subculturing for a while, co-evolution began with the RNA of the host and the RNA of the parasite, and new parasites were born.

Specifications of the system that produces parasites

There is room for parasites to occur in this system (parasites seem to almost always appear) because there are vulnerabilities in the replication process that can be used by others. In this system, RNA does not replicate by itself, but has a structure that allows an RNA replication enzyme to replicate itself. RNA with a sequence ★ recognized by the replication enzyme as the replication target is replicated. The parasite is RNA of sequence ★ that lacks a gene that encodes a replicating enzyme, and because of this missing gene, the parasite's sequence is shorter and therefore it has a faster replication rate than that of the host RNA.


Specifications of the system that produces parasites

This does not depend on the specific implementation target, so it can be installed in any kind of system including software simulators. In particular, Professor Ikegami's Machines and Tapes system seems to be easy to embed because the model structure is similar.

Multi-phased self-replication process

The multi-phased process automatically removes the difficulty of self-referencing.

Explanation that evolved from RNA world to single cell

This research / experiment is, as it is, an explanation of how the RNA world has evolved into a single cell.

Since RNA needs a cellular structure to make and retain a replicating enzyme, it becomes a prerequisite for self-renewal. From that starting point, it already is the simplest form of a cell and easily developed into a fully functional ancestral cell.

Summary of "Asymptotic burnout and homeostatic awakening: a possible solution to the Fermi paradox?"

By the authors, Michael L. Wong and Stuart Bartlett

In this paper, we explore the potential connections between the emergence of cities and globally connected technological civilizations and the Fermi Paradox. With this perspective, we explore the hypothesis that planetary civilizations may inevitably face a self-induced catastrophe that we call “asymptotic burnout” caused by their underlying social dynamics that drive the superlinear scaling of key metrics, including total energy consumption. Civilizations that develop the capability to understand their own trajectory will have a window of time to affect a fundamental change to prioritize long-term homeostasis over unyielding growth, avoiding burnout via a consciously induced trajectory change that we call “homeostatic awakening.” We propose that the longstanding Fermi paradox may be explained by the inevitability of civilizations to either collapse from burnout or redirect themselves to prioritizing homeostasis, a state where cosmic expansion is no longer a goal, making them difficult to detect remotely.

For more details, check out the paper at

Bringing Physics to Life, an Ideas Lab from the Templeton Foundation

by Lana Sinapayen

This event held in Czech Republic brought together early career scientists from 3 big fields: Artificial Life, Biological Chemistry, and Nonequilibrium Physics. It was an intense workshop where we had one week to write several grants to apply for up to $5 million in total funding to tackle a research question related to Origins of Life research. Thankfully, the event was highly structured along a tried and tested schedule that saw us generate dozen of ideas individually or in teams, and gradually select the most promising to develop them through several cycles of writing and presentation. Without giving away any details, the topics broadly ranged from planetary science to information theory, alternatives to the RNA-world hypothesis, and new definitions of life. It was exhausting but rewarding. I left having expanded my network of colleagues through true collaborative work rather than simple conference-style chitchat, and with many new ideas to advance my own research. The organization was top notch, the discussions between different fields both difficult and enriching. I would highly encourage anyone to apply to relevant "Ideas Lab" if they have a chance.

Upcoming Deadlines:

Call for Volunteers