A good example in the US is cigarette taxes - while originally designed to fund cancer treatment and anti-smoking initiatives, a lot of states have come to depend on the revenue from those taxes. As more people switch to e-cigs, which are not taxed and offer big cost savings, states are talking about taxing e-cigarettes similarly.
If the goal of the cigarette tax was harm reduction, presumably there's some incentive to have people smoke e-cigarettes instead of regular tobacco - but the real concern is the loss of the tax revenue the state has come to depend on. Ringfencing (theoretically) discourages this.
I do a lot of work with my local council to deliver better child safe cycle infrastructure. The Local Sustainable Transport Fund is coming to an end in 2016. Cycle City Ambition Grant money is an ad-hoc grant that only impacts 8 cities and many transport ideas would take years to deliver.
Providing a national "fund" that councils could put together projects and then "bid" for funding with a focus on delivering healthy transport options (walking/cycling/public transport) would enable the delivery of huge long-term health benefits.
One of the reasons London does so well with some of the cycling schemes is that there is a committed 10-year plan to invest in cycling. Without this type of financial commitment you don't get a good safe network. You get little bitty pieces of ad-hoc infrastructure that the council hopes to connect together eventually IF future grants become available.
Also, hypothecation doesn't really seem to me to rise above the level of a PR labelling exercise -- if sugar tax income is all spent on NHS obesity programmes, then the government will simply reduce the amount of NHS funding from the general pot correspondingly. Further, you get awkward effects if peoples' behaviour really does change where obesity programmes could end up underfunded because the sugar tax income falls. Better to just fund them to the right level from the general pot, I think.