Why the Same Physics Ideas Get Found Again and Again

In the paper's words

Independent discovery of identical mathematical structures (polygenesis) and the parallel emergence of similar physical behavior in unrelated systems (convergence) suggest a common organizing principle. Existing work identifies fixed‑point structures in scalar‑curvature geometry, cosmological attractors, dual‑subgradient optimization, strong‑randomness renormalization, nonequilibrium functional RG, hierarchical glass models, and quark confinement, yet lacks a quantitative link between the geometry of theory space and the likelihood of independent discovery. We model theory space as a metric flow on the set of Lagrangians, define a Euclidean similarity metric, and derive two complementary quantitative results.

The abstract, verbatim; every line below is quoted from the text.

What it claims

that polygenesis probability scales as N^-2 (basin‑volume model) or as the Herfindahl concentration (alternative model)—would be falsified if a systematic historical analysis revealed a flat or increasing rate of independent rediscovery for models with larger N, or if observed coincidence frequencies for a well‑characterized basin distribution deviated from the predicted P_k values.

Numbers it reports

1. Assuming polygenesis probability is proportional to basin volume, $$ P_{\rm poly}(N)=\kappa\,V_{\rm basin}(N)=\kappa\,1.0258\,N^{-2}. $$ Setting $\kappa=1$ for a dimensionless scale gives $$ P_{\rm poly}(50)=1.0258\times 50^{-2}=4.103\times10^{-4},\qquad P_{\rm poly}(60)=1.0258\times 60^{-2}=2.849\times10^{-4}. $$ The ratio $P_{\rm poly}(60)/P_{\rm poly}(50)=0.694$, a $30\%$ reduction when the required efold number increases from $50$ to $60$.

   from 4.3 Linking basin volume to polygenesis probability

2. Pairwise coincidence : $$ P_{2}= \sum_{i=1}^{5} p_{i}^{2} =0.1600+0.0625+0.0400+0.0100+0.0025=0.2750. $$ Triple coincidence : $$ P_{3}= \sum_{i=1}^{5} p_{i}^{3} =0.064000+0.015625+0.008000+0.001000+0.000125=0.088750. $$ Dominant‑basin share : $$ \frac{p_{1}^{2}}{P_{2}}=\frac{0.1600}{0.2750}=0.581818\approx0.582. $$ Odds ratio between dominant and runner‑up basins: $$ \frac{p_{1}^{2}}{p_{2}^{2}}=\left(\frac{0.40}{0.25}\right)^{2}=2.56. $$ These numbers illustrate how the Herfindahl concentration of basin volumes directly determines polygenesis frequencies.

   from 4.4 Alternative Herfindahl‑based polygenesis model

3. Herfindahl polygenesis model. $P_{2}=0.275$, $P_{3}=0.08875$, odds ratio $2.56$ (Section 4.4).

   from 5. Results

4. Assuming a modest literature‑mining effort requiring $10^{4}$ USD in researcher time and database access fees, the projected cost of the polygenesis audit is $O(10^{4}\,\text{USD})$, with uncertainty dominated by the completeness of the bibliographic corpus.

   from 7. Practical Implications and Engineering Pathways

How it is done

1. We model theory space $\mathcal{T}$ as the set of all admissible Lagrangians $\mathcal{L}$ equipped with a metric $d(\mathcal{L}_1,\mathcal{L}_2)$ defined by the $L^{2}$ norm of the difference of their functional coefficients after a common field redefinition: $$ d(\mathcal{L}_1,\mathcal{L}_2)=\sqrt{\sum_i (a_i-b_i)^2}, $$ where $\mathcal{L}_1=\sum_i a_i \mathcal{O}_i$ and $\mathcal{L}_2=\sum_i b_i \mathcal{O}_i$.

   from 3. Methods

2. Using $N_{\min}=50$ and $N_{\max}=70$ we compute $$ \frac{2}{N_{\min}} = \frac{2}{50}=0.04,\qquad \frac{2}{N_{\max}} = \frac{2}{70}\approx0.0285714, $$ $$ R = 0.04-0.0285714 = 0.0114286. $$ Treating the basin as a 2‑dimensional disk ($k=2$) gives $$ V_{\rm basin}= \pi R^{2}= \pi (0.0114286)^{2} \approx \pi \times 1.3061\times10^{-4} \approx 4.103\times10^{-4}. $$ Normalizing to $N=50$ yields the constant $$ C = V_{\rm basin}(50)\times 50^{2}=4.103\times10^{-4}\times2500=1.0258, $$ so the explicit scaling law is $$ \boxed{V_{\rm basin}(N)=1.0258\,N^{-2}}. $$

   from 4.1 Basin‑volume scaling from inflationary attractors

3. Assume an illustrative landscape with five basins: $$ p_{1}=0.40,\; p_{2}=0.25,\; p_{3}=0.20,\; p_{4}=0.10,\; p_{5}=0.05, $$ which sum to $1$.

   from 4.4 Alternative Herfindahl‑based polygenesis model

4. Simple method: $1000$ ($\epsilon=10^{-3}$) and $1\,000\,000$ ($\epsilon=10^{-6}$).

   from 5. Results

What would falsify it

1. We set $k=2$ because $n_s$ and $r$ span the observable subspace; inclusion of additional observables (e.g., non‑Gaussianity) would increase $k$ and steepen the basin‑volume decay.

   from 6.1 Limitations

2. The central claim—that polygenesis probability scales as $N^{-2}$ (basin‑volume model) or as the Herfindahl concentration (alternative model)—would be falsified if a systematic historical analysis revealed a flat or increasing rate of independent rediscovery for models with larger $N$, or if observed coincidence frequencies for a well‑characterized basin distribution deviated from the predicted $P_{k}$ values.

   from 6.2 Failure modes and falsifiability

3. Falsification criterion (basin‑volume model).

   from 6.2 Failure modes and falsifiability

Where the claim stops

1. The $L^{2}$ metric on coupling coefficients assumes linearity and neglects possible curvature of theory space; alternative metrics could modify the $N^{-2}$ scaling.

   from 6.1 Limitations

2. The derivation assumes a uniform prior over $N$; realistic model‑building practices may favor certain $N$ ranges, altering $R$.

   from 6.1 Limitations

3. The illustrative five‑basin distribution is an assumption (A1); real theory landscapes may have different basin fractions, affecting $P_{k}$ values.

   from 6.1 Limitations

Where to read

1. Abstract209 words · 1 min
2. 1. Introduction261 words · 1 min
3. 2. Background and Related Work312 words · 1 min
4. 3. Methods311 words · 1 min
5. 4. Analysis272 words · 1 min
6. 5. Results83 words · 1 min
7. 6. Discussion341 words · 1 min
8. 7. Practical Implications and Engineering Pathways264 words · 1 min
9. 8. Conclusion135 words · 1 min
10. Version History22 words · 1 min
11. Provenance and AI Involvement225 words · 1 min