# Possible attack / bug though using the distance rule

**URL:** https://forum.duniter.org/t/possible-attack-bug-though-using-the-distance-rule/643
**Category:** Discussion
**Created:** [4 January 2016 14:35 UTC](https://forum.duniter.org/t/possible-attack-bug-though-using-the-distance-rule/643 "2016-01-04T14:35:45Z")
**Posts on this page:** 1
**Showing post:** 6

<div class="post-metadata">

### Author: ![cgeek](https://forum.duniter.org/user_avatar/forum.duniter.org/cgeek/32/279_2.png) [@cgeek](https://forum.duniter.org/u/cgeek)
#### Post date: [6 January 2016 14:24 UTC](https://forum.duniter.org/t/possible-attack-bug-though-using-the-distance-rule/643/6 "2016-01-06T14:24:29Z")

</div>

I meant we don’t need your initial “90%” rule neither the rule of counting only the 2 ways certifications.

We could add the 4) recommandation, maybe under the form of a:

> It is a good practice to certify the people who certify you, both for courtesy and web of trust security.

Or something alike. 🙂

> [@Arcurus](#):
>
> is the calculation efficient enough to calculate it on the fly, or should it be reevaluated every time frame similar to how currently memberships are handled?

Yes it would be handled a similar way: each time frame (i.e. when a new valid block is received) we compute for each member if he is a _sentry_1 or not.

It would not be _that_ costly because we do not need to make the complete computation of the WoT each time, we just need to do it on members impacted by:

- broken links (just expired certifications)
- new links (new certifications)
- new members
- gone members

* * *

[1] that’s how I call the members against which we test the WoT rule

---

_[View the full topic](https://forum.duniter.org/t/possible-attack-bug-though-using-the-distance-rule/643)._