The methodology paper
TargetSpace: Benchmarking Personal Intelligence by Target-Specific Forecasting Under Partial Observation.
Abstract
Many systems claim to understand specific targets, yet almost none can show their forecasts are target-specific rather than reflections of base rates and routine. TargetSpace makes that distinction measurable: prospective, sealed, proper-scored forecasting of target-state transitions under partial observation, certified by an own-routine baseline (R2), a calibration gate, and a permutation specificity test, with an evidence-tier ablation and an architecture-neutral grid. Its novelty is the conjunction assembled around one question — is the forecast about the target, or about the average? — not any single ingredient. Personal world modeling is the flagship instantiation; the framework is a shared, multi-track apparatus. The work is a pre-pilot proposal: it reports no empirical results and ships only a synthetic demonstration harness.
Positioning and using TargetSpace
A summary of the paper's positioning section.
A complementary, under-measured axis
TargetSpace is not the "best world-model benchmark overall." It addresses a distinct, under-measured axis — target-conditioned longitudinal world modeling — that physical-plausibility, generation-fidelity, and control-success benchmarks do not, by construction, score.
Compared on shared axes only
Physical reasoning, video generation, embodied robotics, symbolic forecasting, and agent memory are complementary families. TargetSpace is compared with each only where they overlap — see the comparison table.
A battery of controls
Apply it as a protocol: zero/short/longitudinal history, shuffled-history and wrong-target controls, modality ablation, oracle and human anchors, calibration, and evidence attribution. See Baselines.
Positioning statement
TargetSpace evaluates a missing layer — whether a model can transform passive longitudinal observation into a target-specific predictive model — central for agents that reason about particular people, teams, systems, or environments over time.
Cite
@misc{sylvester2026targetspace,
title = {TargetSpace: Benchmarking Personal Intelligence by
Target-Specific Forecasting Under Partial Observation},
author = {Sylvester, Yuri Andrade},
year = {2026},
note = {Preprint; pre-pilot proposal, no empirical results},
url = {https://targetspace.org/}
}