The landscape of extremely low-resource machine translation (MT) is characterized by perplexing variability in reported performance, often making results across different language pairs difficult to contextualize. For researchers focused on specific language groups—such as ancient languages—it is nearly impossible to determine if breakthroughs reported in other contexts (e.g., African or American languages) result from superior methodologies or are merely artifacts of benchmark collection. To address this, we introduce the FRED Difficulty Metrics—Fertility Ratio (F), Retrieval Proxy (R) Pre-training Exposure (E) and Corpus Diversity (D) —that serve as dataset-intrinsic metrics to contextualize reported scores. Our findings reveal that a significant portion of result variability is explained by train-test overlap and pre-training exposure rather than model capability. Additionally, we identify that underperforming XLR languages—particularly extinct and non-Latin indigenous languages—suffer from poor tokenization coverage (high token fertility), highlighting structural limitations of transfer learning for languages outside pre-trained models’ representation space. By providing these indices alongside performance scores, we enable more transparent evaluation of cross-lingual transfer and provide a more reliable foundation for the XLR MT community.

Cuneiform is the oldest writing system used for more than 3,000 years in ancient Mesopotamia. Cuneiform is written on clay tablets, which are hard to date because they often lack explicit references to time periods and their paleographic traits are not always reliable as a dating criterion. In this paper, we systematically analyse cuneiform dating problems using machine learning. We build baseline models for both visual and textual features and identify two major issues: confounds and distribution shift. We apply adversarial regularization and deep domain adaptation to mitigate these issues. On tablets from the same museum collections represented in the training set, we achieve accuracies as high as 84.42%. However, when test tablets are taken from held-out collections, models generalize more poorly. This is only partially mitigated by robust learning techniques, highlighting important challenges for future work.