A real problem at the current time is the "final weekend big MTF". It's decisive and it sells the race to the sponsors, but is all too often a damp squib, because either the GC is sorted by then and the climb is neutralized, or the riders are too afraid of the huge final climb and they don't attack. Bola del Mundo was a success in 2010 because the race was close, but did nothing for the race in 2012; Angliru in 2013 and Zoncolan in 2014 did nothing other than add an unexpected name to the list of victors there (both from the break, which reduces a bit of the lustre of the climb). Ventoux 2009 can go in the same file.
The Giro has historically been the best at knowing how to do the final mountain showdown properly, but then, the Giro has the best options for doing so, with the Mortirolo-Aprica, Gavia-Tonale and Finestre-Sestrière options. And even then, if the GC is already sorted you get a tame stage (see 2011). There aren't so many options for this kind of stage in the Vuelta or Tour, plus there is the tendency to sell the big mountaintop finish. After all, what was the most hyped stage of the 2009 Tour? Ventoux. It wasn't the superbly-designed Le Grand Bornand stage, the best stage Christian Prudhomme has ever greenlighted.
Looking back the last few years, there are two things that have really harmed GT design.
1) The final week of the 2011 Tour being very good, and also the success of Europcar meaning massive home interest. The 2011 Tour was hopelessly backloaded, and while there were some interesting stages in the first half of the race, none of these had any real GC impact other than to crash people out; because nobody had lost any time, everybody still had something to protect, so nobody was willing to yield in the péloton and everybody was nervous. This pattern repeated in 2012, 2013 and 2014 whilst also giving total backloading as a successful formula, because we got an exciting climax, with Schleck's long-distance raid, Contador's aggression on the 109km Alpe stage and the GC being settled on the penultimate day in the TT.
2) Contador's ban being backdated and Alejandro Valverde crashing out of the Vuelta, meaning that the 2012 Vuelta was a big duel between the three top Spanish names of a generation; two of which were specialists in short-to-medium length steep slopes, and one of which won the race with an epic raid from distance in a stage you wouldn't have expected it. It became one of the most successful Vueltas with fans, resulting in Guillén repeating the formula, despite the fact he was extremely fortunate to capture lightning in a bottle thanks to a fortunate set of circumstances - it was a very good candidate for the worst designed GT of all time, barely covering a third of the country, looping back on itself constantly and featuring over half the stages ending on steep ramps of some description.
What the races need is to mix things up. The 2015 Vuelta is, believe it or not, a step in the right direction, although it has many, many flaws. The Tour will always have more of a problem in respect of the pacing of its mountain stages, owing to the mountains being in very specific areas of the country (and their reluctance to use the Massif Central with any degree of difficulty) and, worse, the areas of one of those mountain ranges willing to pay all being very close to one another, leading to repetitive and dull parcours, over climbs that all the riders know like the back of their hand.
The issue is less about using the various climbs, it's more about using them smartly. The 2011 Giro was otherwise quite poorly designed, especially for a Zomegnan special. But it did a few things right. We shall leave out the neutralized stage and talk of such things, for they were not part of the planning of course. They had an early, relatively easy mountaintop that sorted the contenders from the pretenders (they used the most famous ever climb for this, Montevergine di Mercogliano), something that all GTs should have an approximation of, because it eases the nerves in the bunch when fewer people are trying to protect something. The penultimate weekend stages were done perfectly. The Sunday stage (the 5-mountain odyssey to Rifugio Gardeccia) would always be in the back of riders' minds and intimidate them, putting them off taking risks to attack the previous day, so the Saturday was an MTF on Zoncolan, a climb that will create gaps simply because of how evil it is. If those stages are the other way round (ignoring geographical constraints) we do not see riders like Rodríguez and Arroyo attacking 3 climbs out, Nibali never attacks on the Giau descent, and everybody waits until the last 5km because they're afraid of the Zoncolan. Similarly, look at the 2009 Vuelta, where the well-designed stage to Velefique was soft-pedalled in fear of Sierra Nevada and La Pandera; then compare to the 2010 Vuelta, where they placed the hardest multi-mountain stage at the tail end of the three mountain stages with the hardest MTF (Lagos de Covadonga) in the middle, with better results, racing-wise. If the hardest MTF has gone, and riders don't have that difficult a finish in the final stage, they can't leave it until then, and then if they must leave it til then, they can't go all in for the final climb; they've got to go earlier. And that leaves us with spectacles like Aprica 2010 or Sestriere 2005.
The other problem is that flat stages simply aren't hard enough a lot of the time to put any pressure on. When a flat stage creates some fun, we've had some great stages - the echelon carnage in the 2013 Tour, for example. Sure, I'm not suggesting they turn it into a total lottery like the Middelburg 2010 slugfest, but while they're trying with the cobbles in the Tour and the sterrato in the Giro, the problem that too many GT contenders are too well-protected in the flat stages is becoming an issue. Before the GC has been sorted out the peloton is too tense and nervous to throw in some real narrow roads like we see in the Classics, so maybe an Ardennes-type stage early on to create some small gaps, then followed by some tough flat stages that test bike handling, strength in the wind and so on would be of use. Or, you know, to break up the order, an early time trial that's of a decent mid-length, like the Cholet one in 2008. Not so long that the time gaps are enormous halfway through week 1, but longer than a prologue and long enough to set the status quo before we get to the first mountains.
And for Christ's sake, it's a Grand Tour. There should be two ITTs of at least medium length, preferably two longish ones.