EC-Bestiary: a catalogue of metaphor-based optimisers

A catalogue, and the argument it is making

The Evolutionary Computation Bestiary describes its own scope plainly. It is a catalogue of the exuberance of the metaheuristic ecosystem, a list of the animals, plants, microbes, natural phenomena and supernatural activities that can be spotted in the metaphor-based computation literature. The background it gives is a history. The field has long drawn inspiration from natural systems, with classics such as Genetic Algorithms and Ant Colony Optimization, and the last two decades brought an explosion of natural and sometimes supernatural heuristics, from birds and bees to zombies and reincarnation. The project is careful about what it is and is not. It says explicitly that the list makes no claims about the scientific quality of the papers listed, and that it is not a hall of fame of algorithms, offering a literary comparison to a place with a few good creatures vastly outnumbered by mindless beasts. For a practitioner, the useful reading is that this is an index and a provocation, not a recommendation engine, and that the entries are deliberately arranged so that a validated method and a metaphor method sit next to each other in the same alphabetical list.

The joke is the argument

The tone is set immediately. The page opens with a quotation from Machado de Assis, whose line about madness being a continent rather than an island frames the whole enterprise, and then the disclaimer makes clear that the authors do not endorse or recommend any of the methods listed, find most of them ridiculous, and consider many a useless waste of space. The passage about submitting a paper is written as a joke, but it carries a real argument underneath: a paper that adds another animal to the literature without doing the analysis deserves the treatment. The framing matters for how you use the resource. A neutral catalogue would be easy to misread as a shopping list, and a sarcastic one cannot be. The authors also state their own view of where the field should go, that the literature could use more mathematics and less marsupials, and that the community should grow past its metaphor-rich phase, which they compare to chemistry outgrowing alchemy. That is a position, stated openly, and it is the frame for every entry that follows.

The reading list: papers that take the metaphors apart

Before the catalogue, the README lists papers that anyone working with metaphor-based metaheuristics should read first, and this is the most immediately actionable part of the project. They fall into recognisable groups. There is the general critique, starting with Metaheuristics: The Metaphor Exposed and a call for action paper on metaphor-based metaheuristics, and a paper on lessons from the Bestiary itself. Then there is the exposure work, covering the grey wolf, moth-flame, whale, firefly, bat and antlion algorithms as six misleading optimisation techniques inspired by bestial metaphors, and a paper exposing the chimp optimisation algorithm as a misleading metaheuristic with structural bias. Then the case studies, a rigorous analysis of the harmony search algorithm explaining how the research community can be misled by a novel methodology, an analysis of why cuckoo search brings no novel ideas to optimisation, and the intelligent water drops algorithm with the argument that it cannot be considered a novel algorithm. And then the comparisons, an analytical and numerical comparison of biogeography-based optimisation against genetic algorithms, a paper asking how novel the novel black hole optimisation approach is, and a paper on the grasshopper optimisation algorithm asking whether it is a novel algorithm or a variant of particle swarm optimisation. If you are about to build something new in this space, this list is the first thing to read, and it is also the bibliography a reviewer is likely to hold you to.

How each entry is recorded

The catalogue entries follow a consistent bibliographic form, which is what makes the collection usable as a reference rather than a novelty list. Each entry gives the algorithm name as a heading, the author list with the publication year, the full paper title, the journal with volume and page range, the ISSN where one exists, and a DOI link. The A section alone shows the range of the entries: the African Buffalo Optimization of Odili and Kahar from 2016 for the travelling salesman problem, the Al-Biruni Earth Radius metaheuristic from 2023, the artificial algae algorithm from 2015, Ali Baba and the Forty Thieves from 2021, the American zebra optimisation algorithm from 2023, a bio-inspired method for the constrained shortest path problem called Amoeba from 2014, the amoeba plasmodium work from 2018, and the Anarchic Society optimisation for PID control from 2012. Names that need a colon for a second variant, as with Amoeba: Plasmodium, are listed as their own entries rather than folded into the parent. The machinery behind the references is in the repository, with a bestiary_references directory, a doi2bib.R script for turning DOIs into bibliography entries, an analysis_files directory, and a README_maker directory that appears to assemble the page.

Validated methods and metaphors in the same list

The most consequential editorial decision in the project is that classics and metaphors share the alphabet. The README states plainly that the Bestiary puts classic and well-validated works of the metaheuristics literature, naming genetic algorithms and ant colony optimisation, and some methods described in mostly metaphor-free language, naming JTF and CFO, side by side with others whose scientific rigor is, to put it mildly, lacking. There is no quality filter, no star rating and no recommended column, and that absence is deliberate. The consequence for a reader is that the list is a map of the field, not a ranking, so the only way to use it responsibly is to treat the presence of a well-known algorithm as a sign that this catalogue includes the classics, and to read the critique papers in the reading list before assuming a novel animal is novel. This is also where the project's usefulness to a reviewer or a grant writer lies. A list of papers with DOIs and journal details, arranged so that a reader can see the clusters, is a fast way to establish what has already been tried and what has already been dismantled.

Keeping a catalogue of animals up to date

The project is a static site, with a Jekyll configuration file, an images directory, and a Cages directory, and it is archived on Zenodo under a DOI that the README badges. The dates tell the story of a curated, occasionally updated resource rather than a living codebase. The page says it was updated on 2026-07-01, which matches the last push to the repository on 2026-07-01, while the two tagged releases are much older, version 2.0.1 and version 2.0.0, both from 2018-06-20. There is no test suite, no package manifest and no build pipeline to speak of, which is exactly what you would expect from a project whose output is a document. The one piece of real tooling is the analysis and reference handling in the repository, where the reference data, the DOI to BibTeX conversion and the analysis files are separated from the rendered page. The licence is the open question here, because GitHub could not classify it, so anyone who wants to reuse the list, fork the data or republish the catalogue has to resolve the terms with the authors first. For a document like this, that is a small but real blocker on reuse.

Editorial conclusion

Adopt EC-Bestiary as a research map if you are about to implement a metaphor-inspired optimiser, want to see the shape of the literature, or need the DOI of the paper that already showed your grey wolf variant is a particle swarm method in disguise. Do not adopt it as a source of algorithms, since the project states it makes no claims about scientific quality and explicitly says it is not a hall of fame, and do not treat its licence as settled, because GitHub could not classify it. Before you cite anything from it, check the DOI against the publisher, since the entries are transcribed from papers and the record was last updated on 2026-07-01 while the tagged releases date from 2018.

Frequently asked questions

Official sources

1. fcampelo/EC-Bestiary on GitHub
2. Issues
3. README
4. Releases