Biomedical Engineering Reference
In-Depth Information
What are the challenges ahead for mobile devices? Considering further the
changing interfaces for an increasing abundance of devices, should we imagine
a time without them? This may be hard to consider for molecules, but in reality
molecules can already be Tweet'ed as SMILES [77] or InChIs [78]. But what
about biology? What other uses can we put such devices to that might be col-
laborative in nature? Time will tell, but ultimately, as mobile devices increase
in computing power, they may become our primary computers and means to
access scientifi c research resources, analytical machines, and data in real time.
The science of biomedical R&D may change in the years ahead and will
more likely involve more crowdsourcing, precompetitive collaborations, and
aggregation of data from diverse sources. Working on technologies for mobile
chemistry and biology applications may be a fruitful outlet for developers,
especially if there is a collaborative component that can leverage network
effects. There certainly needs to be more consideration of how these small
powerful computers can be used, and that in itself is a challenge and may come
about as much by accident as by design.
28.8 CROWDSOURCING TAIL FOR
COLLABORATIVE DATABASES
Large-scale Internet systems (such as Twitter, Digg, Wikipedia, Amazon,
Netfl ix) show a long tail in which a few people participate a lot and a lot of
people participate a little [79]. Chapter 6 describes how major company efforts
at crowdsourcing ideas [similar to Innocentive (www.innocentive.com)] like-
wise follow a power law [80] which is most often the signature of a system with
positive feedback [81]. We have investigated whether different types of crowd-
sourcing environments requiring data contributions behave in a similar manner.
For example, we have examined the data regarding the number of uploads of
data in CDD (www.collaborativedrug.com) for each user and the number of
depositions and curations for ChemSpider (www.chemspider.com) users. The
CDD data suggest a power law with a considerable downward tail (Fig. 28.4),
which is a signature of “saturation” of the audience; that is, in a fi xed universe
of users of these software, a majority of possible people are becoming active
contributors (e.g., uploading data or adding content). In addition, the slope of
the apparent power law (solid line) is quite different from the narrow range
typical of much larger scale corporate intranet challenges (dashed line and
Chapter 6) which are usually from 2.7 to 3. The ChemSpider data also show
power law relationships with even more different slopes (Fig. 28.5).
The slope of a power law is a measure of contribution by whom. ChemSpider
content is very strongly driven by a small but very active minority (one to fi ve
persons) of the total audience, while in contrast, corporate intranet challenges
depend most on a large number of occasional contributors, with CDD in
between. Because these statistical signatures are robust and affect our strategy
to engage present and future contributors, this is an important area for inves-
Search WWH ::
Custom Search