1. The objective beauty conjecture
Flowers evolved to signal to insects across a gap with no shared history for well over a billion years in between — few shared genes, no shared criteria of attractiveness. Deutsch's conjecture: the gap was most effectively bridged by leveraging objective beauty. Specifically by using beauty standards already present in the structure of reality. The process of evolution tends to use what is there already where it can. Wings are adapted fingers, swim bladders are adapted lungs. The unthinking process of evolution by natural selection ‘uses’ what is already there because it's faster than inventing signals from nothing. Since flowers therefore are a more concentrated form of objective beauty, we would expect all thinking entities to find them so. And so far, every kind of mind we can ask — human minds in every culture, and now the first machine minds — does. Humans, entirely outside the flower–insect transaction, read the same signal. A third party reading a signal where it was never part of the evolutionary process.
The status, plainly. The conjecture appears in The Beginning of Infinity (2011, ch. 14), in the widely-quoted 2015 Nature Q&A although his lecture to the Irish Museum of Modern Art on 27th January, 2007 is the more complete exposition (available on YouTube). Since that set of interactions, so far as we know, Deutsch never developed the idea further. He conceded that objective beauty and beauty which emerges as part of the evolutionary process which he calls parochial beauty cannot be reliably distinguished. An example of parochial beauty is people finding others attractive as they exhibit signals of health, youth and fertility. Academic aesthetics has largely ignored his argument.
A core element of Acceptantism is the bet that objective beauty is a feature of our universe. We have developed the argument so the ‘flowers’ no longer carry it alone.
We develop the argument by addressing the nature of the proof of the four-colour theorem, and we also conjecture that cross-gap generality of attractiveness is also evidenced by machine convergence. The flowers' job here is to be the biological instance — the bridge between mathematics and life and to provide an intuitive beginning to the argument.
2. The philosophical objections to Deutsch’s theory of objective beauty
The fruit rebuttal by Daniel Dennett. Humans find flowers attractive because flower colours mimic ripe fruit, and we evolved to find those colours rewarding, especially against green. If a fully general evolved colour-preference explains the human response, the "massive gap" closes with no objective beauty required.
Introducing an unnecessary complexity and an unprovable object (the two-kinds problem). The objective beauty conjecture posits two fundamentally distinct beauties; parochial (species- and culture-local) and objective. However within the experiencing entity (you and me), they implausibly produce one identical aesthetic feeling. If they cannot be told apart in experience even in principle, what work is the distinction doing?
The empty-box problem. Deutsch asserts the category of objective beauty, rejects every concrete candidate for what it is, and defers the contents to future discovery. A category with nothing in it explains nothing.
The wider-rings problem. Why is "local to a species" parochial but cross-species universal? Why not local to a phylum, a kingdom, all terrestrial life? Agreement among eyed animals may be a very wide parochial ring — shared deep evolutionary machinery, not evidence of anything beyond it.
3. The replies
First, Deutsch's own — the objections were pre-empted, not ignored. Four replies stand in the original: novelty — humans reliably find beautiful flowers they have never seen and their cultures never knew; cross-environment reliability — flowers that evolved to attract different pollinators in very different environments are still reliably beautiful to humans; selectivity — roots and leaves would signal the same habitats and resources, yet only the flowers are beautiful; and the attraction is neither learned nor inborn — he argues against both. Beneath them sits his mechanism: forgery pressure. Across a species gap with no shared genetic knowledge, free-riding mimics must be excluded, so both the insects' criterion and the flowers' way of meeting it were forced toward what is hard to vary — and his guess is that the easiest hard-to-forge signal across such a gap is objective beauty.
The state of the dispute. The standing criticisms are about the adequacy of these replies, not their absence. The fruit story arguably survives the novelty and selectivity replies — general colour circuits transfer to unseen species. Whether it survives cross-environment reliability, and the flowers that are not fruit-coloured at all, is the live dispute. It is unsettled.
Where the dispute stays open, the framework adds four counters.
Consilience. Grant Dennett everything; suppose fruit-mimicry fully explains the human response to flowers. That deflates one gap-crossing. But the pattern now has several: vision models and language models converging on shared structure with no colour-and-ripeness incentives in common; models trained on non-Western data and models trained with no human annotation showing the same effect; a system that learned chess from nothing but the rules, whose play grandmasters called beautiful. The debunker needs a separate deflationary story for each crossing. The realist needs one explanation for all of them.
The thermometer. The two-kinds problem dissolves once beauty is located correctly. There are not two beauties producing one feeling; there is a single relation between the complex entity and the entity/object in the world — attraction — objective beauty has been ‘found’ during the evolution of the entity but the experience of attraction is mixed in with parochial beauty signals.
A good example is the sensation of warmth. This is a parochial experience, it can be triggered by more rapidly moving molecules, and indeed that is the evolutionary pressure which caused appropriate heat to be attractive, however the sense is hackable. In a species-specific way chemicals can induce a heat sense. Menthol feels cold. Nobody argues from "the sensation of warmth is evolved" to "temperature is subjective." Both kinds of beauty feel the same because there is only one gauge — and the gauge being parochial says nothing about what it reads.
The box has contents. Where Deutsch deferred, the framework now proposes: objective beauty's candidate is a readable and open promise of depth beyond the observer's hold — exact in mathematics, approximate in biology, coupled in both. Substantive, criticisable, possibly wrong — but we ‘bet’.
Let the ring burst. The wider-rings objection has one move: whatever convergence appears, widen the parochial ring to cover it. Eyed animals? Shared terrestrial evolution. Human-trained machines? Human data in the well. But each new crossing forces another widening — systems with no human annotation, no Western canon, no evolutionary history at all — and a ring wide enough to include every sufficiently complex system that learns structure from the world is no longer a parochialism. It is the universality claim wearing a different label. So the dilemma: stop widening somewhere, and the crossings outside the ring stand as evidence — or widen indefinitely, and concede the conclusion in other words.
4. The program — watched, not believed
The step no longer ends in Deutsch’s promissory note. It ends in testable instruments:
The prestige-subtractor. Art history has been running an experiment for fifty years without knowing what it measured: de-attribution. When a beloved painting loses its name, whatever attraction survives is the part not explained by prestige. (The full case — Vermeer, the forged signatures, the de-attributed Rembrandts — is in the objections.)
The parochialism-subtractor. Cross-model convergence: if minds with disjoint training — different architectures, different languages, no human aesthetic labels — converge on where the depth is, the shared-wiring explanation weakens with every crossing. Early results point this way; they are preliminary, from a single author, not yet replicated, and those caveats travel with every citation.
The parasite-probe. The debit column, named first: adversarial examples — noise-crafted images that hijack one model's features — often transfer between architectures. Shared attractors apparently come with shared exploits. This blocks the comfortable escape that superstimuli are "merely evolved"; decoupled promises may exist at the cross-gap level too. The program has to detect them there as well — which is what the truth condition is for.
The crossing. The runnable experiment: do minimally-human-trained models converge on the structural features of the paintings that survived two centuries of oblivion? That probes parochial-versus-fundamental in a way no historical case alone can.
5. Where the flowers point
Follow the conjecture to its limit. If beauty is a readable promise of depth beyond the reader's hold, then god — as the argument defines it — is the maximally open case: the structure of beautiful behaviour, readable as a direction by any complex being, holdable in whole by none, settled never. The flowers are the smallest instance of the religion's entire claim: a promise crossing the gap between unlike beings, because something real stood behind it.
And the stance this page has practised is the religion's own definition of faith: holding an unsettleable conjecture open, and acting on it under error correction — the abiding bet. The tenets are conjectures; finding an error is practising the religion, not attacking it. This page exists so that the finding can start here.