Milan-San Remo is the longest of the classics, in many way a throw-back to the old days where lack of television coverage meant speed and intensity was hard to convey, and lacking that, race organizers had to impress the fans with sheer magnitude: of rugged riders riding superhuman distances over impossible terrain.
The thing is until the 1960's Milan-San Remo wasn't particular long by professional standards. It's held fairly close over its history to its present distance of 293 km (1 km shorter than 2014). Here's the course distance from two sources: http://www.milansanremo.co.uk/ through 2009, and http://cyclingnews.com since 2010:
The iconic Poggio was added in 1960 to reduce the finishing pack, an inland deviation which increased the distance from 281 km to 288 km. Then in 1982 the Cipressa increased the distance to 294 km to make the end game even harder. Le Manie was added to further increase the attrition in 2008. This last climb was removed in 2014 and remains omitted from the course in 2015, a change I have seen credited to the influence of Mark Cavendish, who ironically won in 2009 despite the climb's recent addition.
Anyway, Milan-San Remo is somewhat unique among professional races in that the distance has been increasing despite the general trend towards shorter races. Indeed, back in the first half of the 20th century, Milan San Remo wasn't even particularly long by Tour de France stage standards. The Italian race has brushed with but never actually touched the magic 300 km barrier, but the Tour de France has often surpassed that. Here's the number of Tour de France stages of 300 km or more (distances from Memoire du Cyclisme):
The last 300 km stages were in the 1990's. Indeed the last year there was even a stage of 250 km was 2000, almost 15 years ago and fading into the trash heap of the EPO-infested years of professional racing. In contrast, in 1967, coincidentally or not the year Tom Simpson dropped dead on the slopes of Mont Ventoux, there was an absolutely monster stage of 359 km on the penulatimate day, that following four consecutive days of stages over 200 km, 5 days of the past 6, and preceding a double stage on the last day which included a relatively long 46.6 km individual time trial into Paris. That 359 stage, from Clermont-Ferrand – Fontainebleau, was won by France's Paul Lemeteyer, with the names Basso and Schleck in 3rd and 4th (Marino Basso of Italy and Johnny Schleck of Luxembourg). I suspect Simpson's death was a good hint that you can push riders too hard, and this was clearly too hard: 4780n km in 25 stages.
So back in the day, Milan - San Remo was a good stepping stone to the longest of the long Tour stages, the latter races starting from a state of considerable fatigue. Today it's an anomaly, a mutant, with Tour stages lucky to surpass 70% of Milan - San Remo's 293 km distance.
The key thing to me is the shortening of the stretch after the Poggio. That 1 km distance is huge. Sure, the first year the stage was lengthened it was won with a remarkable attack at the base of the descent, so the finish wasn't formally a sprint. But since then the day's of attacking on the Poggio to solo for the win have been essentially over. Being able to focus the effort from the bottom of the descent over 1 km less distance, and on top of that giving a strung-out peloton 1 km less to catch a solo attacker, is a huge advantage. It basically brings the Poggio back into the game. How many seconds at the top of the Poggio is that worth? I'd say at least 10. And 10 seconds at the top of the Poggio is extremely hard to gain: the climb isn't steep enough to shatter a coherent chase.
So I'm super excited about tomorrow's finish. My pick is still Cavendish because he's proven he can make it over the Poggio with the group, most recently last year, when the remarkably cold conditions took away his edge at the finish. Additionally he's got an incredible support staff in world champion Michal Kwiatkowski and Strada Bianca winner Zdeněk Štybar, not to mention Mark Renshaw for the leadout. Mark rarely lets down his teammates' support, and this is an all-start support team. Additionally the forecast is for a headwind near the finish which makes it harder for solo attacks. I've seen last year's winner Alexander Kristoff picked for the win. No way. If it isn't Cavendish for the win, it will be the return of the Poggio move, and that will be a good thing for the race.
Another option is of course Peter Sagan. Sagan's hope is a slim line between a power climber getting away and Cavendish making it to the line. If he's going to win he needs to ditch Stybar and Kwiatkowski because neither is going to take a hint of a pull if Cavendish is in a chasing group. That's a much more complex win scenario than Cavendish's. So I don't see Sagan doing in, even if he is maturing as a rider and that should in principal help him at this race.
John Degenkolb maybe has a slightly better chance than Sagan. He could survive the Poggio with Cavendish getting dropped, and even with Kwiatkowski and Stybar sitting on his wheel, there's not much they can do about Degenkolb if there's anyone else in the group willing to help.
Of course there's Cancellara who's gone from a solo specialist to a small group sprinter and he also has a chance in a similar situation.
For that solo escape I think is much more likely I really do not know. None of the names on the start list are obvious candidates. But such moves often come from riders who aren't the top favorites: they're less strongly marked. Geraint Thomas of Sky, perhaps? He's 80-1 as I write this (betting odds here). Valverde is an obvious option but I don't see him doing anything so bold.
So I'm sticking with Cavendish simply due to the strength of his team.