One man can't make a difference

At least not directly. But 3.5% of the population can, as long as they take direct, non-violent action. Note that this study focuses exclusively on regime change, and not campaigns in general.

Overall, nonviolent campaigns were twice as likely to succeed as violent campaigns: they led to political change 53% of the time compared to 26% for the violent protests. This was partly the result of strength in numbers. Chenoweth argues that nonviolent campaigns are more likely to succeed because they can recruit many more participants from a much broader demographic, which can cause severe disruption that paralyses normal urban life and the functioning of society. Overall, the nonviolent campaigns attracted around four times as many participants (200,000) as the average violent campaign (50,000). 

Chenoweth admits that she was initially surprised by her results. But she now cites many reasons that nonviolent protests can garner such high levels of support. Perhaps most obviously, violent protests necessarily exclude people who abhor and fear bloodshed, whereas peaceful protesters maintain the moral high ground. By engaging broad support across the population, nonviolent campaigns are also more likely to win support among the police and the military – the very groups that the government should be leaning on to bring about order...  but some relatively large nonviolent protests also failed, such as the protests against the communist party in East Germany in the 1950s, which attracted 400,000 members (around 2% of the population) at their peak, but still failed to bring about change. 

In terms of the specific strategies that are used, general strikes “are probably one of the most powerful, if not the most powerful, single method of nonviolent resistance”, Chenoweth says. But they do come at a personal cost, whereas other forms of protest can be completely anonymous. 

Regarding the “3.5% rule”, she points out that while 3.5% is a small minority, such a level of active participation probably means many more people tacitly agree with the cause... Ultimately, she would like our history books to pay greater attention to nonviolent campaigns rather than concentrating so heavily on warfare.

An interesting and important message. It does focus on sheer numbers very heavily though. It'd be nice to look more in to some of those exceptional cases where large numbers failed and small ones succeed. It's easy enough to see why this would happen in the case of military coups; there are countless historical examples of larger forces being defeated by smaller ones. It's not at all so obvious why protests fail and succeed. It's hard to believe it all boils down to the total fraction of a population. It would also be interesting to consider the much more general conditions under which protests of any kind succeed and fail. There won't be exact rules, but I'd bet anything you like there are common trends.

The '3.5% rule': How a small minority can change the world

In 1986, millions of Filipinos took to the streets of Manila in peaceful protest and prayer in the People Power movement. The Marcos regime folded on the fourth day. In 2003, the people of Georgia ousted Eduard Shevardnadze through the bloodless Rose Revolution, in which protestors stormed the parliament building holding the flowers in their hands.