This much is clear: The coronavirus is becoming more transmissible. Ever since the virus emerged in China, it has been gaining mutations that help it spread more easily among humans. The Alpha variant, first detected in the United Kingdom last year, is 50 percent more transmissible than the original version, and now the Delta variant, first detected in India, is at least 40 percent more transmissible than Alpha.

What’s less certain, however, is how the virus’s increased transmissibility will affect the pandemic in the United States. Alpha’s arrival prompted worries about a new surge in the spring, but one never came. The proportion of Alpha cases kept going up, but the total number of cases kept going down. People got vaccinated. Alpha became dominant in the U.S. Cases fell even further. The virus had become more biologically transmissible, but it wasn’t transmitting to more people.

There was one notable and confusing exception: In April, Michigan experienced a spike in cases that experts believe was indeed fueled by Alpha. The fact that the variant had such different consequences for Michigan than it did for the rest of the country shows just how difficult it is to make predictions. Alpha has little effect on vaccines, but fears about the variants that slightly erode vaccine protection, Beta and Gamma, have also quieted; neither is causing significant case spikes among the vaccinated. “If there’s ever a time that we needed to be humble, it’s around this issue,” says Michael Osterholm, an infectious-disease epidemiologist at the University of Minnesota.

Delta has gotten so much attention because it has the most troubling collection of traits yet: It is markedly more transmissible than Alpha, can sicken a large proportion of people who have had only one dose of a vaccine (though not those who have had two), and may even cause more severe disease. All of this is enough to be a warning, especially as Delta is now responsible for 10 percent of U.S. cases and rising. But as with Alpha, which was also suspected to be more severe, how the variant ends up behaving in the real world will depend on more than its biology. It will also depend on how we—the virus’s hosts—choose to behave, how many more people we vaccinate, and, to some extent, how lucky we get.

Read: Coronavirus variants have nowhere to hide

All of these factors are likely to have played a role in the Alpha-associated springtime spike in Michigan. According to cellphone mobility data from that period, people in the state had gone back to nearly pre-pandemic levels of movement, says Emily Martin, an epidemiologist at the University of Michigan. The Alpha variant also got to Michigan relatively early, and happened to find its way into groups of young people who were not yet eligible to be vaccinated. “It was sort of bad timing,” Martin told me. If Alpha had arrived a little later, or the vaccines a little earlier, then Michigan might have looked more like the rest of the country, where immunization was able to blunt Alpha’s impact. In the race between variants and vaccines elsewhere in the U.S., vaccines won.

Two concepts about viral spread help explain why timing and chance make such a difference. First, the coronavirus spreads exponentially, which means that even a slight delay in mitigation efforts can lead to dramatically different outcomes. Second, the virus’s spread is what epidemiologists call “overdispersed,” which means that the majority of patients do not infect anyone else but a small handful might infect dozens of people. In other words, most sparks of infection do not catch fire. But occasionally a single infection might cause an early super-spreader event, which ends up seeding a major outbreak. “Looking from state to state, it can be like, ‘Well, why is this state doing well versus that state?’ Sometimes it’s just luck,” says Adam Lauring, a virologist at the University of Michigan.

In predicting how variants will behave, much of the world has looked to the U.K., where an excellent and comprehensive genomic-surveillance program has tracked the rise of Alpha and now Delta. Alpha made up 98 percent of all COVID-19 cases in the U.K. at that variant’s peak in March; Delta has since taken over, accounting for almost all new cases. It’s too early to say whether the U.S. will follow the same trajectory. Alpha was responsible for anywhere from 38 to 86 percent of all new U.S. cases last month, depending on the state. Nathan Grubaugh, an epidemiologist at Yale, says this fact suggests the limits of comparing the two countries. “The U.S. is far more heterogeneous than the U.K.,” he told me, with more diversity in viruses and bigger geographic differences in vaccine uptake. When it comes to Delta, he said, “that means some places are going to be impacted harder.” And most likely, those places are going to be the ones where fewer people have been vaccinated.

Experts agree that vaccines are the best way to stop Delta. Data from the U.K. suggest that one dose of the Pfizer vaccine offers only 34 percent protection against the variant, while two doses provide 88 percent. Large swaths of the U.S., however, are still struggling to get people to take any doses at all. A recent Washington Post analysis found more than 100 counties where less than 20 percent of the population has been vaccinated. “Whatever cracks that we have in our program for getting communities vaccinated, that’s what Delta is going to exploit,” Martin said.

The U.S. at least has an ample supply of vaccines to fill those cracks; the challenge is a social one. In countries without enough doses to stop it, Delta portends far more trouble. In some ways, it’s a luxury to say that Delta’s eventual effects in the U.S. are unpredictable.