I have been saying for a year
that the NERC CIP standards, in their current prescriptive format, are unsustainable.
Until my last post
my number one reason for saying this was that a large portion – perhaps even
half - of the effort that NERC entities have to expend in order to comply with
CIP goes toward activities that have no security benefit.[i]
In my opinion, instituting a non-prescriptive, threat-based approach to CIP
would be one way to increase the portion of CIP spending going to security,
without requiring a net increase in spending to achieve this result.
In saying this, I always referred to
“compliance paperwork” as by far the largest (but not the only) component of
this “non-productive” effort. In other words, my proposed solution to CIP’s
unsustainability problem would result in a large reduction in paperwork, although
it wouldn’t eliminate it, since some compliance paperwork would still be
required.
However, the problem with this argument was
that I had to admit there is no good way to tell, simply by looking at a
particular paperwork activity, whether it is “good” paperwork – which
contributes to security and thus would be retained under my proposal – or “bad”
paperwork, which doesn’t contribute to security at all. Given this, an entity
would have no objective criterion for determining how much of their CIP
compliance effort contributes to security; they would just have to take a
guess, based on their experience. So I was basing my argument on something that
might be called an “inherently unverifiable” fact: This is a fact that can
never be proven true or false.
In my last post, I demoted this reason for
CIP’s unsustainability from Number One to Number Two. You can read about my new
Number One reason in the post already cited, but in short the reason is that
the prescriptive CIP requirements force entities to allocate their cyber
security spending (both spending of dollars and “spending” of employee time) to
activities that provide less security benefit – and often much less – than
activities they would otherwise prioritize. In demoting the previous Number One
reason to Number Two (but still saying it was a valid reason), I was in effect
saying that, even if an entity’s priorities for cyber security would – if CIP
were suddenly made non-mandatory - align exactly with the activities mandated
by CIP v5 and v6 (of course the chance of this happening is zero), they would
still be wasting a lot of effort on activities that had no effect at all on
security.
Last week, I spoke in front of the CIP users’
group for one of the NERC Regional Entities about the problems with CIP and my
tentative proposal to fix them.[ii]
There were a lot of really good questions, and we had a great discussion, in
which I probably learned a lot more than my audience did.[iii]
During this discussion, someone expressed
skepticism that any CIP compliance paperwork has zero security value; after
all, documenting what you do is a good practice – and often required for
internal audit purposes – in any activity related to computer systems and
networks. I at first replied with my standard answer described above, that
there is no way that, simply by looking at a paperwork task, an outside
observer could determine that it did or didn’t contribute to security; only
longtime compliance or cyber security staff members at the entity itself could
make this determination – and that would only be based on gut feel. So this
determination will always be inherently unverifiable.
But as soon as I said this, I felt quite uneasy.
This was perhaps because, during the week I made this presentation, there was a
raging debate in the national press about whether the idea of “alternative
facts” was a valid one, or just another way of saying “lies”. And here I was
going one step further by asserting that certain facts were true but just could
never be verified. If the person who invented the phrase “alternative facts”
had instead asserted my concept of “inherently unverifiable” facts, she might
not have received all the flak that she encountered – if anything, the members
of the press would have started looking through the literature on epistemology to see if “inherently
unverifiable facts” might be a valid concept (i.e., can there be a fact that
could never be verified? It’s an interesting question. It actually is a big
debate in physics today, where proponents of string theory, and also the idea
that there are an infinite number of universes, readily admit that these ideas
can never be definitively proven true or false).
I was really not comfortable continuing to
assert that there is no way to identify paperwork that is required for
compliance but doesn’t contribute at all to security. But then I realized there
is no reason to continue to make this assertion, since the result is virtually
the same - whether these activities don’t contribute at all to security or
whether they do contribute but only minimally. The result in both cases will be
that a lot of the paperwork required by CIP contributes very little to security.
So let me stipulate from here on out that every activity required by CIP
contributes in some way to security, although often in a very small way.
Once I admitted that, I realized my Number
Two reason why CIP is unsustainable had now gone away and been subsumed into Reason
Number One, without requiring that I change how I articulate that reason at all.
As I said above (and in my last post), the Number One problem with the CIP
requirements is that they cause entities to use their limited cyber security
budgets to carry out security mitigation activities that would otherwise have a
very low priority – if the entity were free to do what it thought was best.[iv]
Since no NERC entity – at least none that I know of – has an unlimited cyber
security budget, this results in the most important cyber threats (based on the
current threat landscape in North America,[v])
going either unmitigated or inadequately mitigated.
To summarize this post, I no longer believe
that there are activities – which I’ve previously called “pure compliance
paperwork” - that are required by the CIP standards but contribute nothing to
cyber security. Every activity required by CIP contributes in some way to
security, but a lot of these activities make a very small contribution. I am
making a proposal that would rewrite CIP to require that NERC entities
prioritize the activities that contribute the most to BES[vi]
cyber security, without prescriptively saying that certain activities are
required, no matter how little they advance the goal of securing the bulk
electric system.
The views and opinions expressed here are my own and don’t
necessarily represent the views or opinions of Deloitte Advisory.
[i]
I based this statement on informal discussions I’ve had with various NERC
entities, not on any sort of formal poll.
[ii]
I prefaced my remarks by pointing out that I am working, with two co-authors,
on a book that will lay out this proposal, among other things. We expect to
have it out later this year.
[iii]
I will probably have another post inspired by this discussion soon.
[iv]
You may cringe when you hear me say that the CIP standards shouldn’t
unnaturally constrain NERC entities from allocating their limited cyber
security budgets as they “think best”. You may point out that a) a lot of, or
even most, organizations still believe that what is best as far as cyber
security goes is to spend as little on it as possible; and b) even if an entity
realizes it must spend a substantial amount on cyber, it won’t necessarily
spend it in an optimal way, due perhaps to a lack of understanding of cyber
security principles and practices.
Both of these objections can be answered by pointing
out that my “proposal” for rewriting CIP will require the entity (or a third
party) to assess its security posture with respect to various security domains
(software vulnerability mitigation, anti-phishing, change control, etc.) and
develop a plan for mitigating the most important deficiencies identified. This
plan will have to be reviewed by a competent outside party, which might be a
consulting firm or the entity’s NERC Region; this process is similar to the one
now mandated by CIP-014. I am currently
leaning toward the idea that the Regions themselves should do this review. I realize
they don’t currently have the manpower to review all of these plans. That will
hopefully change, but even then the Regions will probably still have to hire
outside resources, at least to address temporary overloads. But since otherwise
the entities would have to engage their own consultants for this task, and
there would be the potential for some consulting firms to go easier on the
entity in exchange for being engaged to do the not insubstantial job of
implementing the mitigation plan (in fact, this is the biggest problem I see
with the PCI standards for payment card security, since the PCI standards are
audited by assessors paid by the retailer being audited, who are then allowed
to be engaged to mitigate the problems that they identify. They have lots of
incentive to downplay the problems in the official report, since they know it
will make the retailer look good), I still think it’s better for the Region to
do it. While having the Regions do it will probably require an increase in the
assessments paid by each entity, the entities will hopefully see that this
simply replaces an amount they would otherwise have to spend themselves.
Having the Region review an entity’s assessment and
mitigation plan will address both of the objections shown above. If the entity
happens to think that their cyber security posture is just great and there’s no
need to spend much more money on cyber, or if the entity’s mitigation plan will
spend too much on unimportant tasks and too little on important ones, the
Region will be able to order the entity to revise all or parts of their plan.
And they will be regularly audited (perhaps even once a year) on how well they
are carrying out that plan.
[v]
My proposal for rewriting CIP – and specifically the one I and my co-authors
will outline in our upcoming book – will require that the team that drafts the
new standards identify the primary cyber threats to the North American bulk
electric system. The entity will be required to address each of those threats
in some way, either to mitigate deficiencies in their defenses that are
identified in an assessment, or to document why a particular threat doesn’t
apply to it. However, since the threat landscape changes very rapidly (e.g.,
phishing came out of nowhere about five years ago to become probably the most
serious cyber threat today, and the origin of more
than half of successful cyber attacks in recent years), there needs to be
some way of continually updating this threat list. I am proposing that there be
a NERC committee which meets at least quarterly to a) assess new threats and
determine whether or not they should be added to the list; b) determine whether
any threats currently on the list should be removed; and c) write and update
guidance on best practices for mitigating these threats.
In addition, since some threats only apply to
particular types of entities or particular regions of the country, there will
always be threats that an entity faces, that aren’t included in the “NERC-wide”
list just described. It will be up to the entity to make sure these particular
threats are also addressed, and it will be up to the NERC Region to verify that
the entity’s mitigation plan adequately addresses these threats.
[vi]
Note that, in my proposal, the CIP standards will still be focused entirely on
BES security. Every NERC entity has other cyber security goals: protecting
financial data, protecting customer information, etc. These also need to be
addressed, but CIP has no bearing on these. In other words, under my proposal
the entity will need two cyber security budgets: the budget to address BES
threats and the budget to address all other cyber threats.
No comments:
Post a Comment