Concentration created at launch doesn't go away on its own
A token's code can be flawlessly written and fully decentralized in the sense that no single party can unilaterally change how it works, and it can still end up functionally centralized because of who holds it. If a small number of wallets end up with most of the initial supply, that concentration doesn't fix itself over time. Those holders retain the ability to sell a large fraction of circulating supply into the market whenever they choose, and because markets for new tokens are usually thin relative to the size of early allocations, even one or two large holders deciding to exit can move the price sharply. Every other holder is left exposed to a risk they didn't choose and can't see coming, since there's rarely any signal before a large wallet decides to sell.
The problem compounds when the token carries governance rights. A protocol can have every line of its smart contracts open source and audited, but if voting power is proportional to token holdings and those holdings are concentrated, then in practice a handful of addresses control protocol decisions: what upgrades happen, how treasury funds get spent, which parameters change. Decentralization at the code level and decentralization at the ownership level are different things, and a launch mechanism that ignores the second one can quietly undo the point of the first. This is why teams designing a token launch treat the distribution mechanism itself as a core piece of the protocol's design, not an afterthought to be handled once the 'real' engineering is done.