Sentiment analysis of the free nutritious meal or Makan Bergizi Gratis (MBG) program on social media X from the perspective of the political economy of development

The Free Nutritious Meal Program or Makan Bergizi Gratis (MBG) is a flagship social development policy introduced by President Prabowo. Beyond nutrition, the program is politically and economically significant because it involves public budgeting, redistribution, local supply chains, Micro, Small, and Medium Enterprises (MSMEs), food safety governance, and state legitimacy. This study analyzes public sentiment on X toward the MBG and interprets the findings from the perspective of the political economy of development. This study uses a descriptive quantitative computational social media design supported by qualitative interpretation of selected posts, rather than a purely qualitative descriptive design. Data were collected using Tweet Harvest in Google Colab with the exact keyword “Makan Bergizi Gratis” for public posts published from January 1 to June 1, 2025. From the 4,200 initially collected posts, 2,774 relevant posts were retained after duplicate, empty, noisy, link-only, repost-like, and irrelevant entries were removed. Indonesian-language posts were normalized, translated into English, and classified using Valence Aware Dictionary and Sentiment Reasoner (VADER) with standard compound score thresholds. Machine classification identified 2,324 positive, 303 neutral, and 147 negative posts. Positive sentiment was associated with nutrition, children, Papua, stunting reduction, and policy support, whereas negative sentiment was associated with food safety, poisoning cases, accountability, and budget priorities. Politically, positive sentiment may strengthen state legitimacy and support budgetary redistribution, whereas negative sentiment exposes implementation risks, fiscal opportunity costs, supplier accountability concerns, and regional contestation over public resources. Because translation-based VADER classification may misread sarcasm, negation, and context-specific Indonesian political vocabulary, this study incorporated a manual validation protocol and treated the results as machine-assisted digital discourse indicators rather than direct measures of national public opinion.