I've decided to take 7 days away from Facebook. I've finally realised that the cost of a free service is only cheap when you don't factor in the irritation value in not having a concrete set of criteria that you pay for, or a minimum agreed level of quality.
Despite my misgivings about G+, I have created a profile, with an incorrect birth date, just to hold namespace, much as I did for Gmail. I will, however, also be taking 7 days away from that, as well.
So this is my only outlet for 7 days.
*twiddle thumbs*
My cousin asked me about moving to a Gmail backed system, given my recent rants about G+ and security, invasion of privacy etc, so I'll talk about why I think it's a bad idea for certain people to move their e-mail into a corporate system like Google's.
Firstly, you have to understand that I'm an old school sysadmin at heart. While I'm happy to use remote services, I've spent too many sleepless nights hacking lockd and debugging NFS, as well as power-cycling misbehaving ATM switches, to ever really trust anything that I can't put my hands on and reboot. Does that mean I don't use these systems? Of course not! Here I am, right now, using a web-based remotely-accessed database with no real guarantees of service.
But am I committing my only copy of my core writings to it? No.
Am I making private notes to myself about my colleagues or students? No, don't be silly. Anybody could be reading this in transit, or stored at the company. How many employees do these companies have? How successful is social engineering? (Answers: A lot, and Very) Statistically, if the data hasn't been sold yet, it will be shortly. Do I care about the stuff I post here? No, it's all public, and under my real name. I have nothing to hide here.
This is not a dig at Google (or, to be precise, Google alone) - this is a sad fact of life. Many security policies fail because someone, somewhere, doesn't do the right thing. In a small business, or locally employed tech support, they know you and you hopefully know them. You have an illusion of control.
If you use Skype, then you may not know that we don't know how Skype is kept secure. This isn't because their mechanism is so amazing that it defeats even the greatest minds - it's because they don't tell us. This is unusual in the security world - or at least the sensible security world - because it's "security through obscurity", the equivalent of strengthening your door lock by putting your own key in among a bunch of 100 others. Good security is based on open, well-documented cryptosystems that tell you (a) if a system is easy to crack and (b) what would happen if you managed to crack it. Right now, I can't tell you if the Skype system is secure or not - I don't know.
But I still use Skype, only I don't use it for confidential matters.
So my idea of trust is potentially a little more segmented in the electronic sphere than that of a normal person. There are things that I will use, although I don't trust them, because I split my life into secure and insecure stuff. The things I trust, I need to know how they are secured; what guarantees I have as to data integrity, protection and security; and I want to minimise the number of opportunities for information to leak. I want the simplest, most secure, mechanism possible.
Once again, I'm happy to use these large-scale 'free' services for certain things, but what I would not be happy to do is to trust confidential work or student data to one of these services because, in the data path alone, I have willingly introduced many points for possible interception AND I'm beholden to someone with whom I have no actual formal business relationship.
The inimitable Sketch put it very well, years ago, on LiveJournal when he said, and I'm paraphrasing, that if you didn't look after your own stuff, you were basically crazy. These days he mostly says "Back up your shit" and he's right again. It's probably slightly disturbing to some people how right Sketch is about these, really quite important, things. I'm not surprised.
As a small diversion, very recently, it was revealed that one of the backbones of internet security, TLS 1.0 (the thing that allows you to run https connections to secure your web connection to your bank, or SSL, if you will), has a vulnerability that allows people to snoop your traffic. Now, if your connection to your mail server traverses your internal company network, you go through a couple of switches, a router that you own, and you're connected. It is a low probability, but not impossible, that someone manages to snoop the right segment and get in the middle for you. They have to get into your network first - which can be easier or harder (yes, yes, Wifi points) but it does decrease your likelihood of getting intercepted.
Now take that out to the internet and go through the 10-20 hops required to get you to the corporate mail servers at 'large mail host X'. Now you've got several times the possibility that someone can get into your sessions. Of course, this requires someone to be in a position to put software analysers in a place where your traffic goes through them because, most of the time, we can assume that reputable ISPs are reliable. So, when you connect through that open WiFi point at the coffee shop, across the Internet, to some free software service that is all about running software - you're not connecting back to the systems that your business runs, which serve your business' aims, which are run by the people you see for birthday cake every week, you're going back to a place where your valuable data is being looked after under some sort of contractual arrangement.
Or, if you're not paying for it, you're sending it to a place where someone is taking money to advertise to you and, by definition, they're sniffing everything you type in order to customise those ads.
But don't worry, because that's the least of your problems. Remember that whole social engineering thing? Your mail is stored on a server outside of your control and there will be a large number of people who can access it. Most of them won't, ever, unless you do something naughty, but, if your business is interesting enough, then maybe some $$$ changes hands and your mail gets opened up.
(There's a reason that my University strictly forbids us to send exams in the mail. And we have an internal staff mail server.)
Why am I picking on the large providers? Because the small local providers have a smaller staff base, more to lose and are less interesting targets full stop. Google is an obvious target for any range of 'boastable' DDOS attacks, it holds the data of a very large number of people with a questionable grasp of data hygiene and security, and it employs a lot of people.
Google will have some incredibly serious policies and procedures to protect your data but, at the end of the day, they can read your data. Even if it takes three people and the Pope to open up the system for reading, and the fact of that event will be written in letters a kilometre high in the Tundra, someone, somewhere will have an emergency mechanism to get around this. Why? Because (a) complex systems often have flaws and (b) sometimes you need to do it, for real, security-related reasons, and the Pope won't return your calls and everyone else is off at the Google Nudathlon. Or, maybe, your friendly neighbourhood government wants to read your data and would really rather that this didn't involve too many people, or require that whole Burning the Permafrost thing. Yes, letters will be written but, in this day and age, what this probably means is that "e-mails will be sent".
And, of course, e-mails can be forged. (For the paranoid among us, we know that we can make this really, really difficult. Hands up everyone who has seen senior management with a crypto fingerprint on their e-mail.)
Now maybe it _can't_ be read. Maybe they are spending the computational effort to encrypt your data in a way that can only be unlocked with your password and they do an on-demand supply of encrypted data across the network, where your local machine, in a grungy browser-based app, to do the very expensive decryption procedure to get your mail. (Of course, they can still get around that by putting in a plug in that reads the decrypted display and send it back, for any one of the browsers you use. Lynx users, you can stand easy. :)
But they're not doing this. Of course Google don't encrypt it - they actively scan your mail for viruses and are ready to hand it over to the government if they need to, so even if they did encrypt it, they do so in a way that they have the keys.
What does this mean?
Right now, if you are in a job where you handle sensitive or confidential personal data (medical or educational are just two) and you move to Gmail, then you had better hope that the following doesn't happen:
1. Someone attacks Gmail to take it down. (Google's too big? Microsoft got taken down by a DNS-related attack recently.)
Why? Because you can't access your mail and do your job because you have put the nuts and bolts of your mail system out beyond the boundaries of your company. There's no reason that this kind of mail system can't be PART of your solution (we have students on it, for example, but not staff, although that's still a little shaky IMO) but this is the worst kind of distributed system: one where you can't get any work done because a computer, that you don't know about in a location that you can't control, has failed (after Lamport).
What would be nice here is local replication and caching - an ability to still do something when the service is away. If you use POP3 or IMAP to bring your mail down locally, your mailbox still exists when you aren't connected. A webmail account is useless with no internet. The price, of course, is that the downloads have to live on a machine - and now you have a store of your data sitting in a valuable, "can be sold in a bar" commodity.
2. Someone goes after your data because they know that you mail juicy or valuable data around.
Why? News Of The World, anyone? If a Prime Minister has recently contracted Horse Herpes, that's news. To a blackmailer, that's money. To the Opposition, that's power. People know that you send exams around for checking? What would a student pay for an exam paper 2-3 weeks before the day?
People are paying several thousand dollars a question for some of the high-end medical and dentistry board examinations. Maybe it's worth $50,000 to a sys admin to read one person's feed. A clever admin could make it part of the virus scanning - all documents have to be scanned, after all.
Yes, personnel checks are thorough and, no, I don't believe any one company is any worse than another. However, people are people. Sometimes bad things happen and people need money. Don't blame me, I'm a believer in solid welfare!
This ignores the many other possibilities introduced because your mail has to go the long way around.
3. Google changes its terms of service and now you have to read all of your mail in Swahili.
Why? If you have a paid relationship with a company, you have some control over the terms of service. If you don't, then welcome to the Facebook fun park. You don't pay for it, they can do what they like. Did you pay for it? No? Then just stop using it if you don't like it!
This is the most amazing deployment of heroin economics - not only is the first one free, but so are all the rest. The only problem is that, once you've got enough mail stored in there, they can do a fair bit to you because the pain of movement is so great. Ad popups? Why not? We're just offsetting our costs. Did we sell your data? Changed the TOS, sorry, didn't we mention that? Closed your account because you had the audacity to use a fake name because you're avoiding your abusive husband? Read the TOS - and we're cutting off your lifeline to your on-line support groups.
4. That whoever moved your company or organisation to Gmail didn't do it with an actual service agreement between your organisation and Google.
Why? Setting up a contract fixes a lot of the problems I've mentioned above because it actually gives you some contractual stick to hit the company with if something goes wrong. This also covers your backside if someone leaks out the confidential data - well, it may get taken into account by the Judge when sentencing, anyway. :)
This is what we did at my Uni. When we moved across to student e-mail, we set up an agreement. Now Google is working, in this space, for us and, apart from the GENERAL problems of having provider outside of our own campus, we've covered a lot of problems.
Look, I've painted a very black picture. Yes, there are problems, but you can deal with most of them by having a sensible password and a contract with the provider. I can choose a good password, but NOT (belated edit) one of my real ones because I don't know who's reading it, but I can't renegotiate my contract in the two spaces that are bothering me at the moment.
My problem with Facebook and G+ is that I cannot reach an agreement with them because of the service I provide. I want to be in there because a lot of my friends are there. But I need to think about whether the cost is worth it and, despite being a free service, the cost goes up every day.
Let me step back and be frank with some alphabetised bullet points.
A. Ultimately, it is very hard to secure any computer system where you ever want to get anything out of it. Some of this is bad software, bad design or bad security, but some of it is just people. Most of the big companies/free services will give you pretty much the same service as local admins - knowing someone doesn't stop them ripping off your shit. Some companies are dodgy, and when they change hands to even murkier people, jump. (Have you noticed that there is a g+.com site that isn't related to plus.google.com? I wonder how many people they've got so far.)
B. Local support, well-trained and properly funded, almost always beats large-scale outsourcing but the key is 'well-trained and properly funded'. The large-scale groups can offer you services and redundant facilities that dwarf anything you can afford, most likely. Yes, there is a risk that network outage will screw you over but you can do things to work around that, often with their help. If you know what you want, and are willing to negotiate and PAY for it, you will probably get a good solution.
C. You get what you pay for in many respects. Free is cheap but it may incur other costs. Worse, your 'free' account may come because you are being sold as eyes or identity somewhere else. Whose domain are you now projecting in your business dealings? Do you have a corporate e-mail or is your work account a Gmail one? What does that say about your organisation? Would you take a large IT firm seriously if they had outsourced all of their e-mail to Google? Should you? What are their reasons?
D. Everything you write online is public. If not now, then some day. Every e-mail. Every comment. Oh wow, you're cleverly concealed behind "RomanticWarrior666". Nice one. I have no idea who you are. Except that you are friends with this person and that person, live in Nova Scotia, went to St Peters Collegiate School, graduating high school in '92, have a birthday on the 3rd of June (but have concealed the year) and your friends call you Kevin. Nice to meet you, Kevin, born '75-'77 or so. Write everything as if everyone is reading. (This may shock some people but I'm pretty much the same, bar swearing, in my work dealings as I am here. That's part of what they pay me for - thoughtful, communicative and passionate. They get the 'crazy' and the 'socialist' for free.)
E. If your data is sensitive, for business or personal reasons, or confidential, then you increase the risk of problems every extra step you take outside of your organisation. Do you want to be the bank who lost 10,000,000 borrowers' records because some meth-head in tech support for XXNet dumped your DB and sold it for $200 because you could move your salaries budget to a different column? Maybe you do, but only because it's not your head that will roll. Lose enough data, they'll bring in extra guillotines. Embarrass your boss sufficiently, and they'll install a special French knitting lady to yell 'Zut alors' as the blade comes down.
F. Facebook is controlling us, we're not in control of Facebook. I cannot choose how Facebook presents my data, as it turns out. Today it's the weird representation. Tomorrow, if I click on an Ad, will my status automatically change to 'Nick Falkner loves Coke!' to free advertise across all of my friends pages? Will I be 'helpfully' auto-subscribed to other people to allow me to 'opt out' if I want to? Even for people I've already rejected? What will the next irritation be?
I don't believe that whole 'boil the frog slowly and it will not jump out' thing, although maybe, now, I should. Facebook isn't evil or trying to take over the world, they're just making changes to their software, which they're giving us for free, to increase their ability to make a profit. That's what businesses do. However, along the way, it's become obvious that they're not making enough money, for them, and that they'll do pretty much anything to make more of it. I realise now that I've been unhappy for months, but I've been slowly boiling and, despite being in hot water, I haven't thought of jumping. Until now.
Right now, I can't predict what Facebook will do next. I don't know what Google are planning to do, but I do know that I don't want to invest too much time in another Google service until I'm sure it's sticking.
So, at least for the next 6 days, you'll find me here. Writing things longer than 500 characters. :)
No comments:
Post a Comment