Grand Theft Auto made him a legend. In July this year workers at Build a Rocket Boy, a video game studio in Edinburgh, were called to an all-staff meeting. Their first ever game, a sci-fi adventure called MindsEye, had been released three weeks earlier - and it had been a total disaster. Critics and players called it broken, buggy, and the worst game of 2025. Addressing staff via video link, the company's boss, Leslie Benzies, assured them there was a plan to get things back on track and said the negativity they'd seen was uncalled for.

Age prediction using brain imaging, such as MRIs, has achieved promising results, with several studies identifying the model's residual as a potential biomarker for chronic disease states. In this study, we developed a brain age predictive model using a dataset of 1,220 U.S. veterans (18--80 years) and convolutional neural networks (CNNs) trained on two-dimensional slices of axial T2-weighted fast spin-echo and T2-weighted fluid attenuated inversion recovery MRI images. The model, incorporating a degree-3 polynomial ensemble, achieved an $R^{2}$ of 0.816 on the testing set. Images were acquired at the level of the anterior commissure and the frontal horns of the lateral ventricles. Residual analysis was performed to assess its potential as a biomarker for five ICD-coded conditions: hypertension (HTN), diabetes mellitus (DM), mild traumatic brain injury (mTBI), illicit substance abuse/dependence (SAD), and alcohol abuse/dependence (AAD). Residuals grouped by the number of ICD-coded conditions demonstrated different trends that were statistically significant ($p = 0.002$), suggesting a relationship between disease states and predicted brain age. This association was particularly pronounced in patients over 49 years, where negative residuals (indicating advanced brain aging) correlated with the presence of multiple ICD codes. These findings support the potential of residuals as biomarkers for detecting latent health conditions.

You might have read the news about how our mail-sorting machines have been dismantled and how our boss replaced our mailbags with a "Flintstones"-style prehistoric pelican that carries your letters in its mouth pouch and says, "Eh, it's a living" every time we put a letter into its bill. And, don't worry, they're all true. We just want to officially state that "voting by mail is easy!" is what we would have said every election up till now. This special year, we've designed a handy guide to help you participate in democracy via mail. You can access this on the Internet.

In a year of High Street gloom one item beamed bright from the sales figures - the giant TV screen. It may still be a tiny segment of the market but sales are rocketing. Dixons Carphone saw a 70% surge in the sale of screens over 65in over Christmas, and a tripling of sales of screens sized 75in or more. John Lewis said sales of 70in TVs have risen 150% compared with last year, partly thanks to a spike in sales from the World Cup. According to the Broadcasters' Audience Research Board (Barb) sales of screens over 40in have been on the increase since at least the beginning of the decade.

Whilst we may trust Netflix to serve us the content we want, or for Google Maps to predict our routes, or for Spotify to recommend us some songs we may like, when we get to work we revert to manual processes and guesswork." They've been using their crowdsourced – but anonymized – data to provide predictive analytics on their AI platform for more than a decade. I'm not that impressed with predictive at darlings Netflix and Spotify. Meanwhile, some enterprises are getting better at predictive. But what Elkinson says rings true. Consumer tech is forcing the issue (miss you, Alexa, I'll be home soon!). Elkington's got a terrific BS detector for AI blowharding. Barb takes AI in a different but equally exposed direction in The value AI brings to marketing. She argues that AI is set to transform marketing – and marketing isn't ready. Demandware's survey found a monster gap between the impact of AI on marketing and what marketers are skilled to handle. Barb talked to Demandware about the story behind the numbers. One key point: the ability to differentiate on the data science and/or algorithms looks to be fleeting – and will last only until tools either commoditize or become mature. The real differentiator, says Barb, will be the data itself – and that data is hard to come by. As she says: "Whoever can get the most and right data is going to win." Yup – I would only add: "Whoever gets the most opt-in data…" It's about your community willingly sharing data for value. If you get that data at the flea market, or through terms of service smoke/mirrors, you're going to lose that edge – as soon as customers figure out you're just another data panhandler shilling their vitals. Jon's grab bag – Stuart wants to know if the UK government is leaving it to Microsoft to handle the digital skllls crisis. When you see "We have painted ourselves into a corner," and "We are what we are," you know Stuart isn't exactly thrilled. Michelle Swan makes her diginomica debut writing about a professional services firm (in the Salesforce ecosystem) that keeps employee turnover to five percent using the weirdest, wackiest metric you could ever think of: employee happiness. It's also about using data to intervene – in a good way – before things go too far down the ol' crudder. Get your media fix with Stuart's The BBC – wanting to be Netflix? I'm with Stuart: don't try too hard to be Netflix. Netflix isn't exactly the master of great original programming either – from that standpoint they are a sub-par HBO. Finally, welcome ServiceNow to diginomica – good timing given the "servitization" of darned near everything. More somber, Ryan Avent's The Wealth of Humans describes the current era of automation and it's threat to human-labor, kicking up a vision of future thick with a jobless miasma."

This paper explores links between basis construction methods in Markov decision processes and power series expansions of value functions. This perspective provides a useful framework to analyze properties of existing bases, as well as provides insight into constructing more effective bases. Krylov and Bellman error bases are based on the Neumann series expansion. These bases incur very large initial Bellman errors, and can converge rather slowly as the discount factor approaches unity. The Laurent series expansion, which relates discounted and average-reward formulations, provides both an explanation for this slow convergence as well as suggests a way to construct more efficient basis representations. The first two terms in the Laurent series represent the scaled average-reward and the average-adjusted sum of rewards, and subsequent terms expand the discounted value function using powers of a generalized inverse called the Drazin (or group inverse) of a singular matrix derived from the transition matrix. Experiments show that Drazin bases converge considerably more quickly than several other bases, particularly for large values of the discount factor. An incremental variant of Drazin bases called Bellman average-reward bases (BARBs) is described, which provides some of the same benefits at lower computational cost.