Phylosaur

About Phylosaur

Phylosaur is an independent educational game about the dinosaur family tree. The goal is to identify a hidden genus by guessing other genera. Each guess shows the deepest named clade that the guess shares with the target, so every attempt narrows the search. The catalogue holds more than 580 genera, from early dinosaurs such as Herrerasaurus to Mesozoic birds.

The name combines two Greek roots. Phylo- comes from phylon (tribe, lineage) and is the root of phylogeny, the study of evolutionary relationships. -saur comes from sauros (lizard), the element used in Dinosauria and in most dinosaur names. The first part also sounds like philo-, from philos (dear, loved).

That double reading is intentional. I made Phylosaur because I love these animals and the way their relationships can be read as a tree, so the name is both a game about dinosaur phylogeny and a small declaration of affection for them. The “lizard” in the name comes from the history of the word; dinosaurs were not lizards.

Phylogenetic foundations

How to read the tree: every branching point is a clade, a group made of one ancestor and all of its descendants. Two genera that share a deeper clade are more closely related than two that only meet near the root. Classifying organisms by shared ancestry, rather than by overall resemblance, is called phylogenetic systematics, or cladistics. Willi Hennig formalized the method in the 1950s and 1960s. Clades are recognized by synapomorphies: derived features inherited from a common ancestor and shared by its descendants.

Dinosauria is a well-supported clade. Features used to diagnose it include a perforate (open) hip socket, a long deltopectoral crest on the upper arm bone and a hinge-like ankle, which together relate to an upright stance with the legs held beneath the body. The group is traditionally divided into Saurischia, with a forward-pointing pubis, long neck vertebrae and a grasping hand with an enlarged thumb, and Ornithischia, with a backward-pointing pubis and a predentary bone at the tip of the lower jaw. Birds (Avialae) are theropods, and therefore saurischians. A group of “dinosaurs” that excludes birds is paraphyletic, which is why palaeontologists say non-avian dinosaurs when they mean the extinct forms.

"The combination of such characters... will, it is presumed, be deemed sufficient ground for establishing a distinct tribe or sub-order of Saurian Reptiles, for which I would propose the name of Dinosauria."

- Richard Owen (1842). Report on British Fossil Reptiles, Part II. Report of the British Association for the Advancement of Science, 11: 60–204.

Game mechanics and educational objectives

Each guess is compared with the hidden genus. The game shows the deepest stored clade the two share, which limits the remaining candidates to that part of the tree. A distant guess may only confirm a broad branch such as Theropoda; a close one can reveal a family or subfamily.

This is a simplified version of what systematists do when they place a new fossil: compare it with known taxa and find where it shares the most derived features. The game replaces character analysis with one fixed, ordered lineage for each genus.

Hints reveal the next clade of the hidden lineage. Once the whole stored lineage is known, further hints give clues about the genus name instead. Two guesses are required between hints. Genera found in completed games enter the Museum, where the Clade Atlas groups them under Theropoda, Sauropodomorpha and Ornithischia. Museum discoveries count toward collection achievements: milestones for major clades, complete families such as spinosaurids and tyrannosauroids, and complete difficulty levels.

• Daily challenges provide the same target worldwide for a selected level during each UTC calendar day.
• Practice mode allows unlimited games without affecting Daily statistics.
• Private challenges let friends race to identify the same hidden genus using a shared code. A rematch starts a new race for the same group.
• The game currently has no fixed guess limit.

Difficulty and public familiarity

There are five difficulty levels based on how familiar each genus is to the public, measured by English Wikipedia pageviews over the last 30 days. Familiar names fall into lower levels and obscure ones into higher levels. This measures public recognition only, not phylogenetic complexity, fossil completeness or scientific importance.

Pageviews are an imperfect proxy: they depend on language, media attention and the collection period. Names shared with plants, places, people or other animals are checked separately, and genera without reliable data are assigned conservatively.

Data sources and taxonomic curation

The genus list started from occurrence and taxonomic records in the Paleobiology Database, followed by filtering and manual review. Lineages combine the project's earlier curated classifications with paths retrieved from The Taxonomicon (S. J. Brands, Universal Taxonomic Services) through the R package taxodist.

Scripts clean placeholder ranks, detect ambiguous names and homonyms, remove playable genera that appear as internal clades, and align older paths to a shared backbone built from well-resolved genera.

Each release is checked for repeated nodes, placeholder clades, inconsistent depths, playable genera listed as ancestors and clades with more than one parent. When an automatic result is unresolved or loses too much structure, a reviewed lineage is kept and the case is logged for later review. The result is a working tree, with a record of where each placement came from.

Museum summaries come from Wikipedia. Illustrations are reusable files from Wikimedia Commons or explicitly licensed Dinopedia pages, checked for subject, file type, licence and quality. Each image shows its creator, source and licence.

Mathematical foundations

Proximity is defined as follows. Let $L_A$ be the ordered lineage of a guessed genus $A$ and $L_B$ the lineage of the hidden target $B$. Both start at Dinosauria and run toward more specific clades. If the two lineages agree on their first $m(A,B)$ entries, then:

The last clade in this shared prefix is the deepest common clade represented in the game. The proximity score is the shared length divided by the depth of the target:

The score answers one question: how much of the target's stored lineage has the guess followed? Because the denominator is the target depth $|L_B|$, the score is not symmetric when the lineages differ in length. Two genera can share every stored clade and score 100%; only the exact name wins.

A related distance counts each parent-child step as one unit:

This value is not shown during play. It counts the steps that separate the two paths after they diverge. Named clades are not evenly spaced in time, anatomy or evolutionary change, so neither $s$ nor $\Delta$ measures geological time or morphological difference.

Each guess works as a query: the revealed clade removes every candidate outside that part of the target path, and hints remove candidates more directly. Branches are very uneven, with hundreds of genera in some clades and only a few in others, so the choice of guess changes how much you learn. That is most of the strategy.

Let $V=\{v_1,v_2,\ldots,v_n\}$ be the thirty-day pageview totals of the catalogued genera. Each $v_i$ is compared with fixed thresholds: Level I at 4,500 views or more, Level II at 1,300–4,499, Level III at 550–1,299, Level IV at 250–549 and Level V below 250. Fixed thresholds keep the meaning of each level stable as the catalogue grows, and an existing genus moves at most two levels per recalibration. These are empirical game categories, not a general measure of difficulty.

Methodological considerations and limitations

Phylogenetic trees are hypotheses. The tree used here is one reading of the available evidence and can differ from analyses based on other character data, taxon samples or methods. Support also varies across the tree: Tyrannosauridae is recovered consistently, while other groups remain poorly resolved because of incomplete fossils or competing interpretations. Even the basic split is debated. In 2017, Baron, Norman and Barrett proposed Ornithoscelida, a clade uniting Ornithischia and Theropoda. The traditional division into Saurischia and Ornithischia is still the most widely used, and it is the one followed here.

Genus names are written in italics; names of higher clades are not. Some included taxa may be nomina dubia, names based on material too incomplete to diagnose, or members of loosely defined groups that are still under revision.

The game needs one ordered path per genus, but published analyses may place a genus in competing positions or leave it in a polytomy, an unresolved branching point. Synonyms are another problem: a genus may be absorbed into another and still remain widely known. Such names may be kept for their educational value, and unresolved placements are logged for review. The tree is detailed, but that detail does not mean every node is equally certain.

Acknowledgments and further resources

Phylosaur depends on the work of the palaeontologists whose descriptions and analyses produced this classification. For further reading: the Paleobiology Database (occurrence and taxonomic records), the Open Tree of Life (a synthesis of published phylogenies), museum research pages and the original description of each genus.

For an introduction to dinosaurs: Dinosaur Paleobiology by Stephen Brusatte and Dinosaurs: How They Lived and Evolved by Darren Naish and Paul Barrett. For phylogenetic methods: Phylogenetics: Theory and Practice of Phylogenetic Systematics by E. O. Wiley and Bruce Lieberman, and Inferring Phylogenies by Joseph Felsenstein.

Phylosaur is non-commercial. Corrections about taxonomy, the interface or the teaching approach are welcome. Please do not cite it as a primary source; trace claims to the original publications instead.