Many pieces of software have a clause similar to:
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
You may recognise that as a large portion of the
BSD 2-Clause open source license. Not to pick on BSD style licenses, section 15 of the
GPL v3, section 11 of the
GPL v2, section 7 of the
Apache license, and a substantial portion of the
NCSA and
MIT licenses also have this clause. The wording is very similar from license to license. Let's look at how this plays out:
- Linux: GPL
- FreeBSD: BSD Style
- OpenBSD: BSD Style
- Bash: GPL
- GCC: GPL
- Clang/LLVM: NCSA
- Apache HTTPd: Apache
- NGinx: BSD
- Postgres: "BSD-Like", Postgres License
- MySQL: GPL
- Apache Struts: Apache
This represents a huge portion of internet-facing devices, including popular services such as
Netflix,
tumblr,
Facebook, and even the
London Stock Exchange. Many devices run
Android, and lots of the routers separating home networks from the Internet are running variants of Linux.
This is not to exclude commercial software, even Microsoft puts these clauses in their
licenses.
I've even come across behaviour in The Co-operative Bank's online banking platform and TSB's Verified by Visa platform (URL structure, exception handling behaviour) that suggest it has a component which uses Apache Struts.
The basic meaning (and I am not a lawyer by any stretch) is that the person or entity who produced the software is (as far as legally possible), in the event of Bad Stuff, they're not Responsible for the Bad Stuff.
So, the story so far:
- Developers kindly write software, entirely disclaim warranty or liability.
- Organisations setup paid-for services off the back of this software, without verifying that the software is fit for purpose, since auditing software is expensive.
- End users then entrust the organisations, via the un-audited software, with their money and personally identifying information (PII).
The end users -- the ones risking their bank deposits, their PII -- are, in many cases (banking has specific protections), the ones who are basically expected to
evaluate the software which they are unknowingly going to use. They are in no position to asses or even understand the risks that they are taking.
With cars, there are safety ratings,
Euro NCAP and the like. Electric goods must meet a minimum safety standard (In the UK, most people look for the
Kitemark) before they can be brought to market. When these goods fail to meet this standard, the seller is held accountable, who in turn, may hold suppliers accountable for failing to supply components which meet the specified requirements. There is a well-known, well-exercised tradition of consumer protection, especially in the UK.
On the other hand, should you provide a service which fails to deploy even basic security practices, you're likely to get, at most, a slap on the wrist from a toothless regulatory body (In the UK, this is the Information Commissioner's Office, ICO). For example, when
Tesco failed to store passwords securely, the only outcome was that the ICO was "unimpressed". There was likely no measurable outcome for Tesco.
Banks, who usually are expected to be exemplary in this field, respond
very poorly to research that would allow consumers to differentiate between the "secure banks" and the "insecure banks". This is the exact opposite of what needs to happen.
The lack of regulation, and strangling off of information to consumers is leading to companies transferring ever larger risks to their clients. Often, the clients have no option but to accept this risk. How many people have a Facebook account because it's the only way to arrange social gatherings (because everyone's on Facebook)? How many people carry and use EMV (in the UK, these are Chip n' PIN) debit or credit cards? Banks are blindly rolling out NFC (Touchless payments) to customers, who have no choice in the matter, and who, in many situations, simply do not know that this is an unsafe proposition, and could never really know.
An illuminating example is
eBay's recent antics. The short version of the story is that a feature for sellers (and hence, a profit making feature for eBay) has turned out to be a Bad Idea, exposing buyers to a very well crafted credential-stealing attack. This has lead to the exposure of many buyer's bank details to malicious third-parties who have then used the details to commit fraud.
In this situation, eBay is clearly gaining by providing a feature to the sellers, and by shifting the risk to the consumer. Further, because this is a profit-making hole, and closing it could break many sellers' pages (thus incurring a huge direct cost), eBay is spinning its wheels and doing nothing.
The consumers gain little, if anything from this additional risk which they are taking on. Like a petrochemical company building plants in heavily populated areas, the local population bear the majority of the risk and do not share in the profits.
This is an unethical situation, and regulatory bodies will normally step in to ensure that the most vulnerable are suitably protected. This is not the case in software & IT services, the regulatory bodies are toothless, and often do not have the expertise to determine which products are safe and which are not.
For instance, both OpenSSL and NSS (both open source cryptographic libraries) are used the The Onion Router (TOR) and the TorBrowser (Effectively FireFox with TOR baked in) are used by dissidents and whistleblowers the world over to protect their identity where their lives or livelihoods may be at risk.
Both
OpenSSL and
NSS have won FIPS-140 (a federal standard in the US) approval in the past. Yet we have had
Heartbleed from OpenSSL, and recently
signature forgeries in NSS. Clearly, these bodies don't actually audit the code they certify, and when it does go catastrophically wrong, the libraries in question maintain their certifications.
For reference, the academic community have been
concerned with the state of the OpenSSL codebase for some time. We've known that it was bad, we've shouted about it, and yet it retained it's certified status.
Individual governments often influence this, by only procuring high-assurance software, and demanding that certain products meet stringent standards, and failures of those products can therefore be financially damaging to the suppliers.
The UK government already has
Technology code of practice which government departments must use to evaluate IT suppliers' offerings. However, there are many more fields which the government has little remit over, and no international remit. The US Government has similar standards processes embodied with the
Federal Information Processing Standards (FIPS, of which FIPS-140, mentioned above, is just one of many).
We also have existing standards processes, like the
ISO 27000 series, which have a viral nature, in that the usage of a external service can only be considered if the organisation aiming for ISO 27000 certification can show that it has done due diligence on the supplier.
However, as mentioned above, these standards rarely mean anything, as they rely on the evaluation of products which very few people understand, and are hard to test. Products that shouldn't achieve certification do, as with OpenSSL.
Clearly, the current certification process is not deployed widely enough, and is not behaving as a certification process should, so we need something with more teeth. In the UK, we have the British Medical Association (BMA), which often takes it's recommendations directly from the National Institute of Clinical Excellence (NICE). If a failure occurs, a health care provider (doctor, medical device supplier, etc.) will end up in court with the BMA present, and may lose their right to trade, as well as more serious consequences.
There is a similar situation in the UK for car manufacture, where cars have a minimum safety expectation, and if the manufacturer's product doesn't meet that expectation, the company is held accountable.
Another example is the US cars being sold into China. Many US cars don't meet the Chinese emissions standards, and hence cannot be sold into China.
We need something similar in software and services: an agreement that, like in other forms of international trade, vendors and service providers are beholden to local law.
We have existing legislation relating to the international provision of services. In many cases, this means that when a company (such as Google) violates
EU anti-competition laws, they are threatened with fines. The laws are in place, but need to be stronger, in terms of what constitutes a violation of the law, and the measures that can be applied to the companies in question.
Currently, internet is the wild west, where you can be shot, mugged and assaulted all at once, and it's your own fault for not carrying a gun and wearing body armour. However, the general public are simply not equipped to acquire suitable body armour or fire a gun, so we need some form of "police" force to protect the general public.
There will always be "bad guys" but we need reasonable protection from both the malicious and the incompetent. Anyone can set up an "encrypted chat program" or "secure social media platform", but actually delivering on those promises when people's real, live PII is on them is much harder, and should be regulated.
Acknowledgements
Many thanks to my other half, Persephone Hallow for listening to my ranting on the subject, and inspiring or outright suggesting about half the ideas in this post, as well proofreading & reviewing the post.
