Smart contracts deployed on blockchain platforms are immutable and control billions of dollars in assets, making their correctness paramount. This dissertation strengthens the Solidity ecosystem through empirically- grounded investigation and feedback-driven tool development. We present the first large-scale mixed-methods study of Solidity’s language design, an- alyzing 296,000 smart contracts, surveying 31 developers, and evaluating 11 analysis tools. Our findings reveal that existing tools detect only 43.5% of error-handling misuses and that documentation instability discourages developer consultation. Building on these findings, we introduce FUZ- ZOL, a syntax-aware mutation fuzzer that discovered 19 compiler bugs (17 fixed), achieving up to 5.8× higher coverage than state-of-the-art fuzzers. We also introduce DGRed, which reduces deployment gas costs by 15.65% through search-based state variable mutation, challenging conventional vari- able packing practices.