Elena Filonova, CIVL Administrator and Competition Coordinator,
writes:

New FS releases were made in order to keep to CIVL Plenary 2018
decisions and to reduce turnpoint radius tolerances forcing all pilots in a
competition to fly the same distance.

In FAI Category 1 events (world- and continental championships), only one earth
model will be used for evaluation:

a. FAI sphere for hang-gliding (2018)

b. WGS84 ellipsoid for paragliding, and from 2019 onward for hang-gliding as
well

The turnpoint radius tolerance (error band) is reduced from 0.5% to 0.05%, with
a minimum of 5 meters.

Important: This rule applies exclusively to FAI Category 1 events, where pilots
can be expected to fly with equipment that complies with this rule.

For FAI Category 2 events, and all unsanctioned events, CIVL recommends

1. Transition to WGS84 where possible (using FS2018)

2. Continue using a tolerance of 0.5%

More info here:
https://www.fai.org/news/fai-sphere-and-wgs-84-fs2016-r16-and-fs2018-r10-are-released?type=node&id=24536

Seems to me the chart should say 0.5% instead of 0.2%.

http://ozreport.com/pub/images/NewFSreleases.pdf

Why do we even need a tolerance?

Calculating the distance between two points, given by their coordinates, is
complicated when assuming that the coordinates represent points on an ellipsoid.
There exist several different ways to do this calculation, some of which are
extremely accurate, but require the high computing power of a PC. Others are
less accurate, but better suited to handheld devices such as our flight
instruments.

To compensate for the slight differences between those different distance
calculation methods, we apply a tolerance. This way, if your vario tells you
that you just barely crossed into a cylinder, but the score keeper's PC, with
its higher computing power and more accurate calculation method, finds that in
fact you did not cross into the cylinder by a few centimeters, you still get
that turnpoint.

A word of caution:

Because of the different distance calculation methods used in FS and in our
instruments, the 0.05% tolerance set by the CIVL Plenary may be too small. To
date we do not know whether the different instruments out there calculate
distance on the WGS84 ellipsoid with sufficient precision to work with such a
small tolerance.

The Paragliding World Cup, for example, reduced the tolerance in 2018 from 0.5%
to 0.2%. For this reason, we currently recommend that FAI Category 2 and
unsanctioned competitions reduce the tolerance - if at all, see above - at most
to 0.2% as well. Once we gain more experience with single earth model
evaluation, we may be able to recommend lower settings.