Promising a privacy-friendly successor to today’s internet, the Safe network is about to go into testing

The internet today is broken (which is why Gigaom recently devoted a series of articles to the subject of fixing it). Control of key elements is becoming more centralized; it is being increasingly censored and manipulated; many aspects are open to intrusion; and the cloud’s need for expensively massive server operations breeds business models that rely on turning everyone’s private data points into dollars.

A bunch of slightly mad Scots at a company called MaidSafe may have the solution. Over an impressively lengthy development period of 8 years, they have come up with something called the Secure Access For Everyone (Safe) network, which they are about to start testing. If it pans out, the Safe network will provide an un-censorable, secure, resilient successor to today’s internet, complete with a built-in economy and the foundations for a distributed, autonomous intelligence.

maidsafe-systemThe idea is not wholly original – a dash of distributed computing here, a soupçon of altcoin there – but it is impressively glued together and actually relatively plausible as these things go. That’s not to say Safe doesn’t have problems, because it does, but they may not be insurmountable.

A post-server world

The Safe network was the brainchild of David Irvine, a network design engineer and serial entrepreneur (his startup Ayrsoft made small business server software called eBoxit) who wanted to improve systems and realized the problem lay in servers. As per the telling of Nick Lambert, MaidSafe’s chief operating officer, servers are an unnatural and unnecessary intermediary in online communications. (“MaidSafe”, by the way, is a play on RAID that stands for “Massive array of internet disks”. Sorry, it’s a horrible name.)

However, there are certain problems with moving away from the server model, as Lambert explained: “If you don’t have servers, what do you log into? You also need to have data that you could almost have with your worst enemy and they still can’t read it and decipher it. The third [problem is] you have to have a network that’s autonomous. The network must be able to heal and manage itself – for a lot of things people are unfortunately the weakest point. They’re very corruptible … this often leads to problems.”

MaidSafe proof of resourceThe Safe model is a bit like [email protected] or perhaps even a botnet, federating the spare capacity of its users’ computers and internet connections to create a distributed supercomputer that obviates the need for centralized servers. Users’ computers, the nodes on the network, can all contribute resources to the pooled effort, including storage, bandwidth and processing power, by running a special app. Data going through the network are broken up into shards of encrypted information that are each stored in at least 4 places at any given time, with no one able to reassemble them but their owners or the intended recipient.

The shared resources are bound together through a routing layer based on the Kademlia distributed hash table, which also underpins filesharing networks like BitTorrent. “It’s fast to the point where, if a node goes offline, the network is aware of it in the time it takes a ping to go, so that’s 20ms,” Lambert said. All this runs on top of the standard TCP/IP internet protocol suite, so it can happily use the internet’s existing hardware infrastructure. “But it is basically rewritten from there up. We’re going from Level 3, the networking layer, right the way up to the application layer.”

A big advantage of this approach is resiliency. According to Lambert, the unintended loss of data on the network would require the loss of power in 4 continents at the same time. And in any case, if that happened, data loss would probably be the least of our worries.

From the user perspective, Safe will potentially provide a similar experience to what people are used to, with a completely different back end. Once they’ve logged in from their desktops, services would come from the network as they come from the cloud today – browsers, app stores, messaging and video chat; whatever developers come up with. “There would be nothing to stop people putting operating systems inside the network,” Lambert posited.

Incentives for users and developers

So how does the network avoid the so-called tragedy of the commons, whereby not enough people contribute resources and a few selfish users chew them all up? Well, people can use the network without contributing, but then they won’t earn safecoins, a Bitcoin-derived virtual currency that exists to reward users (“farmers”) and app developers (“builders”) for their efforts. The network pays out safecoins to developers according to how much their applications get used (direct micropayments from users are also possible) and to users according to their contribution, which is calculated in real time.

MaidSafe app developmentAt the start, safecoins will be quite low-value: when MaidSafe raised $6 million through a safecoin sale in April, it was at a rate of 17,000 safecoins to one bitcoin (worth about $450 at the time). Nonetheless, the outfit hopes Safecoin will become a big cryptocurrency, and that will require a way to exchange safecoins for bitcoins or litecoins or, you know, real money like dollars and euros. Hence, one of the first applications needed on the network will be a distributed currency exchange, perhaps a bit like OpenCoin’s Ripple.

One question that will hopefully get cleared up during the tests, for which 500 developers have signed up, is how quickly a user can farm safecoins. “Unfortunately we don’t know until we get the test networks up and running,” Lambert said. “It’s very much dependent on how much resources are on the network. If there’s lots of people farming, there are fewer chances to earn safecoin. There are so many variables that it’s difficult to predict with any certainty what users can expect.”

A more fundamental problem is the shift to mobile. It’s all very well for a PC user to leave their machine on 24/7 in order to earn as many safecoins as they can, but you simply can’t do that with today’s mobile technology and data pricing. The connections and local processing power can probably handle it, but the batteries can’t – there’s a reason phones are forever going to sleep – and data usage caps are too restrictive. Down the line these things may change, but for now they mean mobile users are only theoretical consumers, not contributors. MaidSafe could pin basic access to the possession of safecoins, but those will be scarce in the early days. It’s a fine line to walk.

So the model isn’t perfect, but it does promise access to effectively unlimited free storage and computing power with anonymity and heavy inherent security, all for donating resources that many PC users can easily spare. The farming/mining system makes more sense than that of Bitcoin, because mining bitcoins involves using a lot of computing power to answer essentially pointless mathematical questions. And the distributed, self-optimizing nature of the Safe network could play very well with both the internet of things and content delivery.

Also, the network is effectively a giant computer that may prove very useful as such – Lambert reckons it could be powerful enough to run an artificial intelligence tat draws on all the knowledge in the network, and in fact MaidSafe has been talking with an EU-funded project called RoboEarth about this very goal.

What’s in it for MaidSafe?

There are effectively 2 MaidSafes:, a for-profit company, and the MaidSafe Foundation, a charity for education and innovation that owns half of the company and will theoretically be funded through dividends once everything takes off.

MaidSafe logoMaidSafe the company will earn safecoins by releasing its own apps, and also by improving the core code. A third revenue stream would come from what Lambert described as a “fairly significant intellectual property portfolio which is there for protection.” Yes, patents – anathema to many in the open source world, but potentially useful for protecting developers on the Safe network down the line.

The product itself is open source, dual-licensed under GPLv3. Anyone can freely use it or even fork it, as long as the resulting product is also open source – if it’s not, then the company requires payment. “It may well be that a centralized current incumbent wants to use some of our libraries; for example, our rUDP and routing libraries may be sufficient for content delivery networks like Akamai(s akam),” Lambert suggested.

In other words, if everything works out then MaidSafe can keep going, and even if the company drops off the face of the earth for some reason, the project code can live on in some future iteration. Either way, I strongly suspect this concept is where a lot of other ideas have been heading. If it’s going to happen, maybe now’s the time — we’ll find out from September, when MaidSafe hopes the initial tests will be successfully completed and the beta phase will begin.

For more information on MaidSafe and the Safe network, read their whitepaper — also check out my recent report on initiatives to reclaim online privacy through decentralization and other means.