On 28 June, 21:45, Terje Mathisen "terje.mathisen at tmsw.no" wrote:
Alan Browne wrote:
It's so easy to **** on the spark of a good idea and so hard to
participate in a worthwhile venture.

Indeed.



"Crowdsourcing" is used in many ways to improve data collection and
analysis, and if there are waters that are not or just marginally
charted, then letting others do it for their personal satisfaction is
entirely valid.
[snip]
Astronomers rely on networks of amateurs to classify galaxies. There are
not enough professionals for the task and they are highly appreciative
of amateur input.

One may need to pay special attention when using charts developed in
this manner, but then all sailors need to heed the sea's harsh lessons
at all times in any case.

The easy way to handle this is simply to use statistical methods, i.e.
when the same area has been mapped by multiple people independently, the
resulting (hopefully!) consensus should lead to both more reliable
mapping and a measure of the quality of the individual mappers.

(Yes, I know that you could in theory generate multiple identities and
supply randomly perturbed versions of the same bogus data, but it will
always be hard to guard against sabotage.)

Here in Scandinavia where we have an extremely fractal shoreline, there
are many, many examples of privately funded buoys and markers.

The most well-known is probably on an infamous underwater rock in the
middle of the fastest path to the Långedrag guest harbor: It resulted in
so many groundings each year that private sponsors paid for a marker to
be put down on it. This marker isn't shown on the official maps, but it
is still very useful.

Terje
--
- Terje.Mathisen at tmsw.no
"almost all programming can be viewed as an exercise in caching"

Hi Terje,
We can use statistical methods once we achieve a high enough density
of data in an area. Beofre that, we have to treat points individually,
and just model the error on those points.
Tim