Decentralize Services:

Start From Email, Social Network, And Maybe Searching

[Revised First Paper]

-- By ShimengCheng - 29 Nov 2012

As an information science major myself, when I graduated from college in 2006, I had never heard any talking about how complex services on the net could restrict human freedom. All I had experienced was the expansion of freedom resulted from the faster-than-ever dissemination of information and the vast knowledge pool open to anyone on the net.

The net itself, at its very beginning (the 60s and 70s), was an autonomous creation invented to break the then monopolizing powers in the telecommunication world – telephone and broadcasting.

I don't think this is true. The designers and builders of the ARPAnet didn't have any such intention, I'm sure: their purposes, to the extent the purposes were clear, were military. And the design of the Net has presumed the presence of the telecommunications network operators from the beginning. Where in the record do you see the signs of such an intention?

For many years following its invention, the packet switching feature of the net has decentralized information dissemination.The packet switching design is itself a decentralized structure. Unlike in a circuit switching system, the communication between two points in a packet switching system does not depend on having a dedicated data transmission path for the duration of the entire communication. At the data transmission level, there is no single entity in the net that is central to the functioning of the whole net. Thus, the net was born decentralized and free.

I don't see this argument so clearly. A packet-switched network could be designed in a server-client, hierarchical structure, as the evolution of the Net in the Microsoft Era showed. The efforts of the Chinese Communist Party or the Iranian Islamic Republic with respect to the Net suggest that a political design of such a kind may also be imposed directly. None of which changes the fact that at layer 1 there are packet-switched networks moving the data, or that the Internet Protocol assumes that packets are the things that get routed in a network.

However, what brings centralization back to the net are the service platforms built on top of the data transmission layer of the net. The existence of service platforms are often justified by the “complicated” functions they serve and the “convenience” brought to the users. The underling assumption is that personal computers or other personal devises cannot economically accomplish these functions on their own. However, the “convenience” provided by service planforms impose cost on human freedom. The service platforms gain enormous centralized power when ordinary functions of the net are performed through them. Under the current net architecture, behavior monitoring through activities on the net can generate enormous unjust profits for email service providers, search engines and social network operators, etc. Decentralizing services requires us to change certain behaviors in our daily life – namely, our dependency on the perceived “conveniences.” Democracy, both in our social structure and in the net, have cost. The “inconveniences” in the net context are just like, in the social context, the cost of everyone taking time off to go to vote in a democratic society. These “inconveniences” will not be a burden to us if we, the netizens, understand their meaningful purposes. It would be great if we have a plug-in personal server like Freedom Box, but before that, we still can do a lot to our emails and social networks by using simple and free softwares.

From an architectural point of view, how can we decentralize the services on the net? Start from our email services: we can each contribute a little storage space in our computer to build our own email server. Since the price of digital storage has become very cheap, it is possible for everyone to set up his or her own email server so that we do not need email service providers. By doing so, we avoid using a centralized platform that is built on top of the decentralized packet switching system. To achieve this, we do not even need a Freedom Box or similar product, we need only a free software that can help us to easily set up our own email server in our personal devises.

A little more is needed. A fixed IP address is necessary. MX records have to point somewhere. Persistent connection is also necessary. But what you say is required, namely storage, is not required. You could build a mail server for yourself on a virtual personal server in the cloud. That would require a few dollars a month. In the long run, it might very well be cost-effective not to own any hardware. The disadvantage is that you have additional legal protection if the storage of your email occurs on physical premises located in the US that you own or control, where the Fourth Amendment applies. But leaving that detail aside, why isn't it simple just to say above that having privacy in one's email requires a few dollars a month?

How about social network? The fundamental components of social network are millions of individual webpages linked to each other. Even before social networks emerged, webpages were frequently linked to one another through the links provided by the webpage owners on their webpages. Then why social network like Facebook has been able to attract millions of users? It is often claimed that the biggest “convenience” offered by Facebook is its “find by name” feature. As long as you know the name of a person, you can friend him/her on Facebook. There is no need to know the web address of anyone's Facebook page. However, if we create our own webpages, put our names on and store the webpages in our own devises, we can easily search the whole web to look for other's webpages – the perceived “convenience” of the “find by name” feature can be replaced by one second of web searching. All we need is a web domain that we have full property ownership, and a free software that helps us to design our webpage.

Once again, a little more is needed. The existing social network of actual friends should not be thrown away just because we are trying to build something secure and private. We want to make migration as easy as possible. And we want to provide people with the ability to make "social applications" easily. That is, we need federated social networking APIs to replace the centralized APIs on which Facebook "apps" now run. This is all possible to develop, but it isn't as simple as you are making out.

How about web searching? Web searching differs from social network and email services in that it is a complicated service that requires constantly scanning of the whole net, massive storage place and good algorithms. The Freedom Box's approach to web searching is to disguise the real search request by randomly generating a few faked search inputs alone with the real search request, so that Google will not be able to keep an accurate log of the real search requests.

No, that's the approach of the existing "Track Me Not" Firefox add-on you could be using already.

This is a smart approach to temporarily combat with Google, but several things may happen following the deployment of this strategy: (1) Google will try to enhance its data mining algorithm so that it can filter out the faked inputs generated by Freedom box's algorithm; (2) Google will need to enhance its search engine to deal with the sharp increase in the number of search requests, since each real search request is now accompanied by a few faked requests; (3) Google may go bankrupt if it cannot get the real log that is critical for its advertisement business.

No, none of this is true, as the history of the Track Me Not add-on shows. The number of additional searches involved is infinitesimal because almost no one uses the add-on. Google doesn't adjust anything. If you are a privacy-oriented web searcher using sensible proxying arrangements, not logging in at Google and using Track Me Not, you are searching with reasonable privacy, and Google doesn't mind, because you are part of the 0.001% of their user base that cares enough to do it right.

If the success of Freedom Box is to completely wipe out Google's business model from the market, then we will be left with no private enterprise willing to offer free web searching.

But neither they nor I nor you expects that to happen. So it's hardly worth worrying about. Replacing Google with federated search will eventually happen anyway, when someone figures out how to do federated search efficiently. You might as well take that remote contingency into account, because it's less remote than the one you are now considering.

In the future when the the data storage become so cheap, the internet connection become so fast and the processors in our personal devises become so powerful, we may be able to each have a search engine of our own in our own house.

Why would you pose the technological alternative this way? If completely decentralized search were possible, federated search would be possible too, and would be obviously better.

But before that, the intermediate solution can be replacing Google with a public search engine funded by tax money, maintained by elected trustworthy engineers, made to disclose important search algorithms and allows individuals to change the search algorithms and to access the stored data. Freedom Box's attack on Google can win us time to develop such a public search engine, so that we can eventually decentralized the power currently owned by the one entity in the net, acquired illegitimately and unjustly.

Search as a public utility is available now, if we want to pay the price for it. But I don't understand why I would want the government to own the search box, even more than I don't want Google to own the search box. And which government in the world is everyone else going to trust to own the search box? Must we then have as many Googles as there are countries that can afford to try to build one?

The approach hasn't solved any problem, and I don't know why it seemed to you as though it would.

The wiki's history facility, which you can find under the "History" button on every page, maintains every version of every topic. By looking at the history, you can see the differences from version to version, knowing exactly who has changed what.

For this reason, you should replace each old version with a new one. If you stack them vertically, the history facility works much less well. Because I use the history facility to understand each student's contribution to the whole wiki, and to study how drafts evolve, I am particularly dependent on your using it as well. I have removed the old draft here. Now I will make a new version, separately, that contains my comments on this draft. You will then replace the contents of the page with your revisions, and the history will work again.