Despite what logic may suggest, switching from regular to diet soda doesn't appear to lead to weight loss -- so proposed "soda taxes" that are intended to socially engineer weight loss are unlikely to have any beneficial effect. Discussion and evidence here.
A 2007 British study used government data on household diets and food expenditures to predict the effect of extending the country's 17.5% sales tax to various foods. When items high in saturated fat were slapped with the tax, the model estimated that deaths would rise by 1,800 to 4,000 a year because consumers would be prompted to switch to foods with more salt.
To fix that problem, the scientists ran a model in which the tax applied to all kinds of unhealthful foods. This time, deaths fell by as many as 2,500 a year, but cholesterol levels rose as people switched from salty foods to fatty dairy items.
And unfortunately, consumers in both scenarios did the last thing any anti-obesity crusader would want: Facing higher grocery bills, they bought fewer fruits and vegetables.
Social engineering on the grand scale rarely works as desired and generally has many unanticipated consequences.