INSIGHTS | PERSPECTIVES
SPILLOVERS AND IMPLICATIONS
The decision by NEAs to abandon technology
may have spillovers on NLAs if NLAs rely on
NEAs to learn about technology or simply follow NEAs’ lead in technology matters. Consistent with the presence of such spillovers,
the percentage of adopters in dorms with a
larger share of delayed NEAs decayed at a
faster rate over time than those with a lower
share (see the graph, center). Results are
magnified when students are geographically
proximate in the same dorm and floor (fig.
S10) or in small dorms (fig. S11).
A final question is whether this behavior also translates to patterns of usage over
time—not just the discrete decision to accept
or reject the technology within the first 2
weeks. In the graph, right, we plot the share
of active users, defined as students who add
new bitcoin over time to their wallets. After
225 days, dorms where the share of delayed
NEAs was above the median share of delayed
NEAs had 45% fewer active users than other
dorms: Higher exit by NEAs corresponds to
lower bitcoin activity by NLAs not just to
Existing literature has stressed NEAs’
positive role in technology diffusion, but
our results highlight a novel, understudied
mechanism through which NEAs might obstruct further diffusion if they refuse to adopt
because their desire to feel unique is challenged or the consumption value they derive
from early, exclusive access is reduced. This is
consistent with qualitative evidence coming
from products such as Google Glass that have
been introduced into the mainstream market
too early, essentially bypassing NEAs.
Many firms strategically manage the avail-
ability of a new technology and rely on a
waiting list to identify NEAs: Individuals
who are more eager to try a new product
sign up first or line up for hours in front of a
store on the day of a release, revealing their
desire for early access. A successful example
of a waiting-list approach is Google’s rollout
of Gmail. The company seeded adoption by
offering Gmail to a few thousand initial us-
ers, who were then allowed to invite the next
batch of users. This transformed Gmail into
a sought-after status symbol. Crowdfunding
platforms have institutionalized the seed-
ing process by capitalizing on NEAs’ higher
willingness to pay for early access to directly
fund the development of new projects (11).
Our results suggest not only that there may
be logistical or financial reasons for not hav-
ing an overly comprehensive launch but also
that, by restricting the reach of a product
launch, managers can fulfill NEAs’ need to
be “unique” and capitalize on their potential
positive spillovers on others’ adoption.
One should be careful when generalizing
our findings, in light of their limitations.
First, we studied diffusion of a unique tech-
nology that has not yet reached mainstream
consumer use, is more complex than many
technologies (12), and is evolving rapidly in
terms of mainstream perception. Second,
although MIT undergraduates share many
similarities with students elsewhere, a subset
of them selects into MIT because of a strong
interest in technology. Third, every partici-
pant had access to the technology, mean-
ing that we measured perseverance with a
technology rather than decisions to adopt
or not. Although students had to invest time
and effort to claim their bitcoin, the distribu-
tion could be viewed as a gift. Our results are
therefore more relevant for cases in which
a firm, government, or institution is trying
to foster adoption of a new technology by
initially giving it away or heavily subsidiz-
ing it. Fourth, we did not have variation in
the length of delay that NEAs experience, so
we cannot give guidance as to optimal de-
lay strategies. The short delay helps us keep
most factors constant (such as the cost of
adoption and availability of complementary
technologies), but makes our results less
likely to translate to settings where delays
may naturally be longer.
Notwithstanding these limitations, the
experimental nature of our data, combined
with fine-grained information on individuals’
behavior, allows us to cleanly estimate the
causal effect of short delays on adoption and
to estimate counterfactual diffusion. When
access to the innovation is visible to individuals who are currently excluded, seeding
a technology while ignoring NEAs’ needs for
distinctiveness and early access is counterproductive and unlikely to generate an optimal adoption cascade. j
REFERENCES AND NOTES
1. Z.Griliches,Econometrica 25,501(1957).
2. E.M.Rogers,DiffusionofInnovations (Simon&Schuster,
3. J.Gans et al.,“Foundationsofentrepreneurialstrategy”
(SSRN Working paper no. 2844843, SSRN, 2016).
4. E. von Hippel,Res.Policy 5, 212 (1976).
5. A. V. Banerjee, Q. J. Econ. 107, 797 (1992).
6. E. L. Glaeser et al ., Q. J. Econ. 111, 507 (1996).
7. D. J. Watts, P. S. Dodds, J. Consum. Res. 34, 441 (2007).
8. S.Nakamoto,“Bitcoin:Apeer-to-peerelectroniccashsys-tem” (2008); https://bitcoin.org/bitcoin.pdf.
9. R. Böhmeetal ., J.Econ.Perspect.29, 213 (2015).
10. J. S. Gans, H. Halaburda,EconomicAnalysisofthe Digital
Economy (Univ. of Chicago Press, 2015), pp. 257–276.
11. A.Agrawal, C.Catalini,A.Goldfarb,in Innovation Policyand
the Economy, J. Lerner and S. Stern, Eds. (Univ. of Chicago
Press, 2014), vol. 14.
12. C.Catalini, J.S.Gans,“Somesimpleeconomicsoftheblock-chain” (NBER Working paper no. 22952, National Bureau of
Economic Research, Cambridge, MA, 2016).
We thank D. Elitzer and J. Rubin for allowing us to study the
bitcoin experiment, the MIT Office of the Provost for support,
and the generous support of alumni. This research was funded
by the MIT Sloan School of Management and the NET Institute.
0 50 100 150 200 250
Days after distribution
NLA in dorm with below median share delayed NEA
NLA in dorm with above median share delayed NEA
0 50 100 150 200 250
Days after distribution
NLA NEA NLA NEA
Dorm Of campus
Not delayed Delayed
Massachusetts Institute of Technology, Cambridge, MA 02139,
USA. Email: email@example.com
Delays discourage early adopters, undermining broader diffusion by later adopters
Delayed access to bitcoin discourages natural early adopters (NEAs) from persisting in more close-knit social environments in dorms but not among those living off campus
(left). Natural late adopters (NLAs) are less likely to persist (center) or to be highly active (right) if they live in dorms that had an above-median share of delayed NEAs.