Coming up Trump: our election model puts Harris behind

U.S. Presidential Election Forecast

• The Economist's model currently assigns Donald Trump a 56% probability of winning, representing a 10-percentage-point shift from mid-August when Kamala Harris held a near 55-45 advantage.
• Trump is ahead for the first time since mid-August, having gained ground in both national polls and crucial competitive swing states despite minimal shifts in raw polling averages.
• The probability shift is driven by two distinct phases where undecided and third-party voters consolidated: first benefiting Harris in August as she captured "soft Democrats," and most recently allowing Trump to convert remaining undecideds and third-party voters.
• While Trump remains less likely to win the popular vote (model suggests Harris retains a 70%+ chance), his gains in the Electoral College have narrowed the gap in key Rust Belt states (Wisconsin, Michigan, Pennsylvania) and Sun Belt states (Arizona, Georgia, Nevada, North Carolina).
• The race has tightened to the point where Trump's support in decisive states has moved from an average deficit of 0.5% to a surplus of 0.5%, effectively flipping the model's win probability.
• Despite the tightness, the model's data editor indicates Trump is currently the slight favorite, stating, "If I had to bet at even money, I would take Trump."

Tesla's Autonomous Taxi Launch

• Tesla unveiled a driverless "Cyber Cab" featuring upward-opening doors, no steering wheel or pedals, and a two-seat configuration, with a target price below $30,000.
• Elon Musk also announced a 20-passenger "RoboVan" and a humanoid robot, though the event provided no concrete details on business models, testing timelines, or regulatory licensing.
• Investor sentiment turned negative post-event, with Tesla's share price slipping 9% amid skepticism regarding Musk's history of missed timelines (e.g., the 2020 promise of a million robo-taxis).
• The industry is highly competitive, with established players like Waymo (700 vehicles in L.A., San Francisco, Phoenix), Cruise, Zoox, and Chinese firms Baidu, Pony AI, and WeRide already operating fleets.
• Competitors primarily rely on expensive sensor suites (lidar, radar, cameras) and significant infrastructure for maintenance and lost-and-found operations, contrasting with Tesla's strategy of using only cameras and a neural network trained on 9 billion miles of data from 6 million vehicles.
• Regulatory approval remains a significant bottleneck, as agencies remain wary of relying solely on camera-based vision systems to handle rare "edge cases" without the transparency of modular sensor stacks.

Microsoft Excel's 40-Year Evolution

• Excel launched in 1984 for the Macintosh and 1985 for Windows, surpassing the market-dominant Lotus 1-2-3 by being the first to recalculate only changed cells and shifting from a command-line to a graphical interface.
• The software has achieved massive scale, with its cloud version supporting roughly 400 million paid users, vastly outpacing its closest competitor, Google Sheets, which serves approximately 10 million paying companies.
• Despite its ubiquity among financial analysts and consultants, Excel has been central to significant data errors, including the miscoding of nearly one-third of early UK COVID gene names due to autocorrect and the inadvertent dropping of health data when legacy row limits were reached.
• Microsoft recently integrated an AI assistant allowing users to perform data analysis via natural language prompts, positioning the tool for AI-enhanced utility rather than obsolescence.
• The program has successfully adapted through the PC, mobile, and cloud eras, with its data editor suggesting AI will make Excel "smarter" rather than replacing it.