226 lines
11 KiB
TeX
226 lines
11 KiB
TeX
%% SCRAP: papers/ANTI_CLAIMS
|
|
%% SOURCE: docs/working/papers/ANTI_CLAIMS.md
|
|
%% STATUS: CURRENT
|
|
%% FITS: ssrn/ch-anti-claims, vol3-research/ch-formal-claims
|
|
%% EDITORIAL: lifted — prose rewritten to press voice
|
|
|
|
\section{Out-of-Scope Claims: What This Work Does Not Assert}
|
|
|
|
This section enumerates claims the StarForth project explicitly does \emph{not} make.
|
|
Its purpose is to bound the scope of peer review: a reviewer who attributes an
|
|
unstated claim to the work is engaging a strawman. Each entry pairs what
|
|
the implementation \emph{does} assert with what it \emph{does not} assert.
|
|
The final summary table (\S\ref{sec:anti-claims-summary}) is the canonical
|
|
reference for authors responding to review challenges.
|
|
|
|
\subsection{Physics and Thermodynamics}
|
|
|
|
\paragraph{Not claimed: a physical theory.}
|
|
The adaptive runtime uses execution frequency as a proxy for thermal energy
|
|
and exponential decay as a model of heat dissipation. These are conceptual
|
|
tools borrowed from thermodynamics, not claims that physical laws govern code
|
|
execution. No actual thermal processes occur in the CPU as a consequence of
|
|
this model; no quantum effects are invoked; energy conservation in the
|
|
thermodynamic sense does not apply to execution frequency.
|
|
|
|
Precise language: \emph{``execution frequency evolves like heat in a cooling
|
|
system,''} not \emph{``execution frequency is heat.''}
|
|
|
|
\paragraph{Not claimed: frequency is thermal energy.}
|
|
The implementation uses a 64-bit integer counter (\texttt{uint64\_t
|
|
execution\_heat}) incremented on each word execution. It does not measure
|
|
joules, calories, or CPU die temperature. The term ``heat'' is a metaphorical
|
|
label. Exponential decay \emph{resembles} heat dissipation and
|
|
steady-state convergence \emph{mirrors} thermodynamic equilibrium in a
|
|
structural, mathematical sense only.
|
|
|
|
\subsection{Optimality and Performance}
|
|
|
|
\paragraph{Not claimed: global optimality.}
|
|
The system demonstrates a 25.4\% performance improvement over baseline under
|
|
the conditions described. This does not imply that no other adaptive runtime
|
|
could perform better, that optimality is proven mathematically, or that the
|
|
implementation outperforms all other virtual machines. The contribution is
|
|
\emph{a working adaptive system}, not the globally optimal one.
|
|
|
|
\paragraph{Not claimed: superiority to JIT compilers.}
|
|
JIT compilers such as PyPy, LuaJIT, and HotSpot share the conceptual goal of
|
|
frequency-based specialization. The claim here is not superior raw speed but
|
|
\emph{deterministic adaptation}: the same adaptive decisions occur on every
|
|
run, enabling formal verification of the optimization mechanism itself. JITs
|
|
typically cannot make this guarantee.
|
|
|
|
\paragraph{Not claimed: universal workload applicability.}
|
|
The system is validated on CPU-bound, deterministic FORTH programs, including
|
|
recursive algorithms (Fibonacci, Ackermann) across workload shapes with Zipf
|
|
exponent $\alpha \in [0.8, 1.5]$. It does not claim to improve I/O-bound
|
|
workloads, non-deterministic programs, adversarial execution patterns, or very
|
|
short-lived processes (fewer than approximately 1{,}000 iterations). Explicit
|
|
failure modes are documented in the companion negative-results section.
|
|
|
|
\subsection{Machine Learning and Artificial Intelligence}
|
|
|
|
\paragraph{Not claimed: AI or machine learning.}
|
|
The adaptive mechanism uses statistical inference (ANOVA, Levene's test,
|
|
exponential regression) and data-driven parameter tuning. No neural networks,
|
|
gradient descent, backpropagation, training datasets, deep learning, or
|
|
reinforcement learning are involved.
|
|
|
|
Precise language: \emph{``statistically-inferred adaptive tuning,''} not
|
|
\emph{``AI-driven optimization.''}
|
|
|
|
\paragraph{Not claimed: the system learns.}
|
|
The system \emph{adapts}: parameters converge to a steady state and feedback
|
|
loops stabilize metrics. This is statistical convergence, not supervised or
|
|
unsupervised learning. Knowledge does not transfer between workloads; the
|
|
steady state is workload-specific.
|
|
|
|
\subsection{Novelty and Prior Art}
|
|
|
|
\paragraph{Not claimed: first adaptive virtual machine.}
|
|
Execution frequency tracking is standard profiler practice; hot-code caching
|
|
is the basis of every production JIT; exponential decay underlies LRU
|
|
eviction. The novelty claim is specific: the \emph{combination} of adaptation
|
|
with 0\% algorithmic variance and formal verification potential, which prior
|
|
systems do not offer.
|
|
|
|
\paragraph{Not claimed: a replacement for JIT compilation.}
|
|
The implementation targets a different niche---verifiable adaptive systems
|
|
for safety-critical contexts---rather than competing directly with LLVM or
|
|
V8. Compilation remains necessary for applications requiring peak throughput.
|
|
|
|
\subsection{Formal Verification}
|
|
|
|
\paragraph{Not claimed: complete formal verification.}
|
|
% PATENT: formal verification claims intersect patent scope; do not strengthen
|
|
Convergence theorems are stated and empirically validated (0\% coefficient of
|
|
variation across 90 runs). Full mechanized proofs for all seven feedback loops
|
|
in Coq or Isabelle are \emph{not} claimed; partial proofs are in progress.
|
|
|
|
Precise language: \emph{``empirically validated determinism,''} not
|
|
\emph{``formally proved correct.''}
|
|
|
|
\paragraph{Not claimed: zero bugs.}
|
|
The test suite comprises 936\raisebox{0.5ex}{+} tests and validates FORTH-79
|
|
compliance. No known correctness bugs exist in the core interpreter as of the
|
|
experimental baseline commit. This provides high assurance, not mathematical
|
|
proof of defect-absence.
|
|
|
|
\subsection{Causation and Interpretation}
|
|
|
|
\paragraph{Not claimed: proven causation.}
|
|
The association between adaptive mechanisms and the observed 25.4\%
|
|
improvement is strong: the functional relationship fits an exponential decay
|
|
model with $R^2 > 0.95$, and only the fully adaptive configuration
|
|
(\texttt{C\_FULL}) improves. No randomized controlled trial establishes
|
|
mechanistic causation.
|
|
|
|
Precise language: \emph{``adaptive mechanisms are associated with 25.4\%
|
|
improvement,''} not \emph{``cause 25.4\% improvement.''}
|
|
|
|
\paragraph{Not claimed: a theory of computation.}
|
|
The work provides a VM optimization technique and a predictive framework for
|
|
adaptive runtime performance. It does not propose a fundamental theory of
|
|
computation, replace computational complexity theory, or define a new model
|
|
of computation.
|
|
|
|
\subsection{Generalization}
|
|
|
|
\paragraph{Not claimed: cross-language generalization.}
|
|
Principles may generalize to other interpreter architectures (Lua, Python,
|
|
JavaScript), but this has not been validated. Compiled languages are
|
|
explicitly out of scope: ahead-of-time optimization already handles hot-code
|
|
without runtime frequency tracking.
|
|
|
|
\paragraph{Not claimed: arbitrary scalability.}
|
|
The system is validated on programs up to approximately 2.1 million word
|
|
executions with dictionary sizes up to approximately 500 entries. Scalability
|
|
to programs with 100{,}000\raisebox{0.5ex}{+} dictionary entries is not
|
|
tested; transition-matrix memory overhead would grow quadratically in that
|
|
regime.
|
|
|
|
\subsection{Deployment Status}
|
|
|
|
\paragraph{Not claimed: production readiness.}
|
|
StarForth is a research prototype suitable for experimental validation. It
|
|
demonstrates feasibility of deterministic adaptation and serves as a platform
|
|
for further study. It is not a production-grade implementation with enterprise
|
|
support or hardening against all security threats.
|
|
|
|
\paragraph{Not claimed: an operating system.}
|
|
The StarForth $\to$ StarKernel $\to$ StarshipOS sequence is a roadmap, not a
|
|
set of delivered products. StarshipOS does not yet exist as a functional
|
|
system; the vision is aspirational.
|
|
|
|
\subsection{Statistical Claims}
|
|
|
|
\paragraph{Not claimed: zero variance in all metrics.}
|
|
The 0.00\% coefficient of variation applies to \emph{algorithmic decisions}:
|
|
cache hit rates and dictionary lookup paths. Wall-clock runtime exhibits
|
|
60--70\% CV due to OS scheduler noise, thermal variation, and cache-line
|
|
effects. These two components are statistically independent (Pearson
|
|
$r = 0.03$, $p = 0.87$).
|
|
|
|
Precise language: \emph{``algorithmic variance: 0.00\% CV,''} not
|
|
\emph{``total system variance: 0.00\% CV.''}
|
|
|
|
\paragraph{Not claimed: 100\% confidence.}
|
|
Results are reported at 95\% confidence intervals. The Bayesian posterior
|
|
$P(H_1 \mid \text{data}) \approx 1 - 10^{-30}$ is effectively certain but
|
|
not unity. Science deals in probabilities, not absolute certainties;
|
|
replication failure, while astronomically unlikely, is not logically
|
|
impossible.
|
|
|
|
\subsection{Scope Exclusions}
|
|
|
|
\paragraph{Not claimed: solutions to undecidable problems.}
|
|
Adaptive convergence to steady state is asserted only for \emph{terminating}
|
|
programs with analyzable workloads. No claim is made about decidability of
|
|
convergence for arbitrary programs.
|
|
|
|
\paragraph{Not claimed: quantum or blockchain components.}
|
|
The implementation uses classical algorithms on classical hardware. No quantum
|
|
superposition, entanglement, distributed ledger, cryptocurrency, smart
|
|
contracts, or neuromorphic computing is involved.
|
|
|
|
\paragraph{Not claimed: a performance record over any named system.}
|
|
No direct performance shootout against PyPy, LuaJIT, or HotSpot is
|
|
presented. The contribution is deterministic adaptation, not a speed record.
|
|
|
|
\subsection{Summary Table}
|
|
\label{sec:anti-claims-summary}
|
|
|
|
\begin{table}[h]
|
|
\centering
|
|
\caption{Anti-claims reference table. Left column: what the work asserts.
|
|
Right column: what it explicitly does not assert.}
|
|
\label{tab:anti-claims}
|
|
\begin{tabular}{lll}
|
|
\toprule
|
|
\textbf{Category} & \textbf{Asserted} & \textbf{Not Asserted} \\
|
|
\midrule
|
|
Physics & Thermodynamic metaphor & Actual physical theory \\
|
|
Performance & 25.4\% improvement & Global optimality \\
|
|
Novelty & Deterministic adaptation & First adaptive system \\
|
|
Verification & Empirical validation & Complete formal proof \\
|
|
ML/AI & Statistical inference & Neural networks or learning \\
|
|
Causation & Strong correlation & Proven causation \\
|
|
Generality & Works for FORTH & Works for all languages \\
|
|
Variance & Algorithmic: 0\% CV & Total system: 0\% CV \\
|
|
\bottomrule
|
|
\end{tabular}
|
|
\end{table}
|
|
|
|
\subsection{How to Use This Section in Peer Review}
|
|
|
|
When a reviewer attributes a claim not appearing in the formal claim table,
|
|
the appropriate response is to cite the relevant paragraph above by section
|
|
number. If the attributed claim does not appear in this section either, it may
|
|
represent a genuine novel claim; in that case, authors should locate the
|
|
supporting evidence in the formal claim table before responding.
|
|
|
|
The intellectual commitment underlying this section is that explicit
|
|
scope-bounding prevents both strawman attacks and over-interpretation by
|
|
readers. Transparency about limitations strengthens, rather than weakens, the
|
|
credibility of the work.
|