Frans, it makes perfect sense to me that a route will change when I change the basic criteria for the choices CT makes, especially with such a restrictive option as 'Routes'. 

However, the idea of maintaining the integrity of an existing route while adding a section with different criteria (for example, a gravel section on an otherwise paved route, or a non 'Routes' option on a long 'Route') already exists. Unfortunately, it involves viapoints. 

If you're not aware, the dialogue box that appears when we click on a VP contains many powerful features. The 'Go' feature might just become a good friend. 

Sorry for responding late, and thanks for your reply.

It does also make sense to me that the route changes. Then again it also makes sense to keep an existing route as is when changing the routing criteria and only apply it to the modifications afterwards.

Of course I can fix it by adding Via points, but that is a bit cumbersome, hence my request.

Frans, 

It does also make sense to me that the route changes. Then again it also makes sense to keep an existing route as is when changing the routing criteria and only apply it to the modifications afterwards.

How does that work exactly? How does CT know when I want it to apply my criteria (Routes, Paved etc.) and not? Am I adding day 15 to my tour or am I diverting to a campsite on day 14? 

Fortunately, we can tell CT when (and what sections) to 'ignore' our routing preferences. That is done using the 'go' feature as highlighted above. One viapoint is necessary. One subsequent click. More info here: https://cycle.travel/help/route_planning/route_types

Frans is wishing to freeze the current solution, except for a bit in the middle (or at start/end) which is unfrozen and to re-plan that on more liberal route type setting.

I can't see how the Go options for Via Points would permit that, but am v happy to be coached. SFAIAA, at present it entails first inserting Via Points to pin the solution arising from the Routes setting, then changing the setting and dragging the route to make the desired deviation or extension. However, on a long route, it could take a lot of VPs to ensure there was no unwanted drift in the section that is desired to stay unchanged. 

It seems to me what Frans is wishing is a solution built up by a concatenation of separate sections. As I understand it, that is pretty much how the software works in finding solutions, except that currently it does not cater for a blend of route type settings. 

So, it would seem conceptually possible for the software to cater for what Frans wishes, something like this: "Starting with a route that has already been saved: freeze solution except between this added special VP and the one that currently is next (if any); allow unfrozen section to be dragged and re-planned according to the specified revised route type setting".  (There would need to be some way of indicating that the revised solution was a hybrid, perhaps by allowing both Routes and Any to be highlighted if the route contains a special VP as above.)

Obvs, that could be a lot of software development, so whether Richard has any appetite to do it is for him; esp considering the perhaps greater benefits from other use of his time (eg doing some cycling!)

Simon, I take my hat off to your mind reading skills. Given that the OP posted no example and was vague in terms of the length of 'shortcut' or extensions I pointed them in the right direction.

The OP stated:

However sometimes I do want to modify the route given, e.g. because I want to use a shortcut or add the part to my start or end point.

There is no problem adding to the start or end point.

In relation to:

Frans wishes, something like this: "Starting with a route that has already been saved: freeze solution except between this added special VP and the one that currently is next (if any); allow unfrozen section to be dragged and re-planned according to the specified revised route type setting".

The following may help:

This is a route (plotted using 'Routes' from Teruel to Valencia. Now, imagine I want to go to Bétera (about 150 km) - I cannot plot a route (to Bétera) because 'Routes' is the active choice.

T Val 1 is the same route, but now I have added 2 viapoints - one at the last point I really want to see on my original route, the other at the next point I really want to see further along my original route.

In T Val 2 I have now added viapoint 3 (between the 2 previous points) and dragged that to Bétera. CT has now calculated a route that keeps me on my original route to my specified point, takes me (using different criteria to 'Routes') to my diversion and then gets me back on my original before I needed.

I'm the designer - do I really need to be reminded that the diversion is not 'Routes'? If so, I can add a note to myself in the viapoint - visible in most gps devices and PDFs.

CT has many powerful tools, and in my humble, this is one of the strongest, especially for those that are very specific about their needs and preferences. It is incredibly versatile. But cumbersome. Apparently. 

No mind-reading, I'm just trying to see the world through someone else's eyes. Most of us don't always express ourselves 100% accurately.

In seeking to replicate your example, I find I can't drag the inserted VP via Betera if the route type remains as Routes. Yet your example seems to achieve it; what do we need to do to permit that?

(I can drag the route if I first change the specified route type to Any (or Paved), but that causes changes in the route; eg around Viver. It's that sort of thing that seems to be what Frans is seeking to avoid, understandably IMO.)

Try using the 'Go & Any' feature on the first viapoint. 

Thanks, I understand now. So CT already caters for what  Frans seems to be after! You just need to know how to do it ...

So CT already caters for what Frans seems to be after! 

Yes. As was said in the very first reply. 

You just need to know how to do it ...

And perhaps to be less adverse to viapoints.