As our city, then our state, then much of the globe started staying at home, we began to compile weekly updates for Segment on the world’s response to COVID-19. The updates were well received, so I’ve compiled the latest research to share here as well.

Physical distancing was the only immediate option, but it’s an extremely expensive way to buy time. Now that we’ve bought time, what happens next?

We need to stop transmission permanently and/or make the disease less dangerous. Overall, the most likely path appears to be a combination of physical distancing and intense testing & contract tracing to reopen as much of the economy as we can until a vaccine becomes available. The least essential activities (like software companies working from shared offices) will likely not return to “normal” until vaccination is nearly universal in 18-24 months.

Stopping Transmission

There are three primary interventions to reduce rate of transmission (R is the reproduction number):


Lead Time



Physical Distancing

2 weeks


Stalls (3x reduction, R ≈ 1)

Testing & Tracing

2 months


Slows (2x reduction, 1 < R < 1.5)


12-36 months


Stops (R << 1)

To stop transmission we need R < 0.5 for an extended period to drive total cases to 0. So far, only vaccines achieve this low of an R.

For physical distancing, without unusually strict adherence it does not reduce R substantially below 1. This means physical distancing is not enough on its own. Two ongoing analyses of great value are monitoring of reproduction number R and Safegraph’s monitoring of shelter-in-place to correlate to R.

For testing and contact tracing, we are seeing signs as of April 6 that Singapore (more here), Hong Kong and South Korea are not able to contain the spread of the virus using testing & contact tracing alone. This is the case even when these strategies are applied with extreme intensity (e.g. forced smartphone bluetooth tracking) that would not be culturally or politically feasible in the US or Europe. Part of the challenge seems to be explained by this paper in Nature showing evidence that peak infectiousness happens in the 2 days before symptom onset.

For vaccines, there is one early vaccine from Moderna in clinical trials in Washington that is hoping to be giving vaccinations to frontline medical staff this fall. However, Johnson & Johnson and others making vaccines seem to think this is very unrealistic and are targeting mouse & primate studies this summer, human safety trials this fall, human efficacy studies in Q1 2021, and then ramping up to 300M-1B doses available by end of 2021, meaning we’re 18-36 months from being able to vaccinate a substantial portion of the world. Maybe 18-24 months in the United States. It’s not yet known what the vaccine rollout would look like, but it’s expected that medical workers and those in high-risk essential jobs would be vaccinated first.

Since physical distancing and testing with contact tracing are the only available interventions today for controlling R, we’re going to need extremely strong testing and contact tracing in order to allow partial reopening while keeping R < 1 for the extended period until a vaccine is widely distributed. Concretely, that suggests: (1) several months of strict physical distancing around R ≈ 1 while testing and contact tracing is put in place, (2) combination of distancing & testing to bring Rt < 0.5 for a while to lower ongoing caseload appreciably, (3) partial reopening with testing & contact tracing for 18-24 months with R ≈ 1 but at low caseload, (4) vaccination to bring R = 0 followed by full reopening in spring 2022. The hope is that the economy begins to recover in 6-8 months when (3) begins, but it is unlikely that Segment returns to the office until (4) is done. California is beginning to show a very similar plan.

Making it Less Dangerous

Alternatively, we could be ok with everyone getting the disease if it was a lot less dangerous. There are two potentials here: asymptomatic caseload and therapeutics.

For asymptomatic cases, there has been much debate about this as a potential “out” because if asymptomatic caseload is very high, then it could mean many of us have already had the disease, it’s much less severe than we thought, and we’re closer to herd immunity than we think. However, as of April 14th Iceland has published data from testing 10% of their entire population. They found only 0.6% were infected and half of those were asymptomatic. China has also published some data that is hard to compare but would put a cap at 75% of people being asymptomatic. This means we are a long way from herd immunity (70%+ required, 50-100x infections so far), and the Infection Fatality Ratio remains in the 0.5-1% range, nearly 50-100x higher than the IFR of H1N1 swine flu. Some studies were recently reported in Los Angeles and Santa Clara counties showing much lower infected fatality rates. However, these California testing results appear to be the result of shoddy analysis. The latest New York State antibody tests indicate similar Infection Fatality Rates. To make matters worse, even mild cases are showing some evidence of permanent lung damage.

For therapeutics, there are therapeutics developed for other diseases that have already been proved as “safe” in humans like antibody plasma transfer, remdesivir, favipiravir, hydroxychloroquine and azithromycin, kevzara, interferon-alpha, umifenovir, lopinavir, ritonavir and others. As of early April most of these are in widespread clinical trials. Most trials will run for several months, and then go into data analysis. So we should have some data by early May or June. These are a crapshoot, I hope we get lucky with an existing therapeutic, and remdesivir and favipiravir in particular are showing some promise, but it’s unlikely to completely solve the issue as overall efficacy is expected to be low (still helpful, but low.) Developing new therapeutics will take much longer: time estimates range from 2-7 years. This is an unlikely path to get us out of the mess.