The issue of cycle-friendly infrastructure presents a significant dilemma. The following is a statement of the position from Active Transport,  ~p.24

1. There are strong grounds for arguing that segregated provision is not the best way to make provision for cyclists in the UK road system.
2. In particular, there is concern that badly designed segregated provision will make cycling inconvenient and unsafe, especially by loss of priority at junctions. Past experience shows that this does not encourage new cyclists but it discourages existing cyclists. The current wording of Rule 61 of the Highway Code is in this regard unhelpful, implying that cyclists must use facilities “unless at the time it is unsafe to do so”. Cycle facilities are rarely safer than the road, and badly designed ones are never safer. It is thus recommended that Rule 61 be omitted.
3. However, segregated provision is wanted by a large proportion of those who are open to persuasion to take up cycling, but have little, if any, personal experience of cycling. This presents a fundamental dilemma. Should public health professionals accept that it is easier to work with the myth and develop segregated facilities in order to encourage people to cycle and hence gain the health benefits? Or should they correct the myth through favourable messages about cycle training, well-equipped road bikes and the directness of the existing road network?
4. Good quality off-road cycle paths, such as those that can be established on old railways or canal tow paths; cycle paths linking quiet streets into through cycle routes; and long continuous quiet routes formed by closing rat runs are popular with established and novice cyclists alike and should be a high priority for cycling investment.
5. There are some roads – generally major rural routes – which are inadvisable to cycle on and where segregated provision is necessary either on the road or avoiding it. It is particularly important that such provision is then funded and implemented to enable cycling. Where such routes would provide direct links between nearby towns or suburbs, they should be a priority.
6. Where the above measures bring into being significant cycle-friendly networks, it would be foolish not to create segregated links to fill gaps in the network (in line with Point 5). These must be of acceptable quality, as advised by experienced cyclists, or they will not be used.
7. The present Hierarchy of Provision should continue to apply. It is the most pragmatic guideline, and has achieved significant increases in cycling levels where it has been applied. The main problem is that too few local authorities have active cycling programmes.
8. It must be stressed that investment in cycling infrastructure will be largely wasted if it is not supported by official endorsement of cycling as a priority mode of transport.

… a risk-averse society is different from a safe society. In a safe society, those who climb mountains take the right equipment, check the weather, ensure that people know their route and expected time of return, know their limitations, and contribute to the funding of mountain rescue teams. In a risk-averse society, people do not climb mountains. Ultimately, a risk-averse society is an unsafe society because people lose the capacity to handle risk sensibly.