Cloud-based solutions only replace one imbalance with another -- and create a whole new set of problems as they do so.
I have always been skeptical of cloud computing. In the rush towards the latest new thing, developers are downplaying problems about privacy and availability. Moreover, the centralization of cloud computing is no more convenient than open source applications that can be installed anywhere you want.
Nor have I changed my mind now that I have tried two examples of cloud-based computing in Google's Chrome OS and Jolicloud, two operating systems designed specifically for netbooks.
In theory, I can appreciate the engineering challenges that both represent, especially for usability, and even make allowances because neither is in final form. Yet, as a user, I feel that both Chrome and Jolicloud have too much of the centralization that I moved to open source to escape. In fact, both seem a step backwards in usability, evoking uncomfortable reminders of the awkwardness of the bulletin boards of the early 1990s.
Currently in beta release, Jolicloud is available as an .ISO or flash drive image. Also in beta, Chrome is less generally accessible, and is best viewed either by compiling from source, or by downloading a VMWare Player image or a VirtualBox appliance. Both run with few problems and are surprisingly stable, considering their stage of development.
Chrome OS: Exchanging the Desktop for the Browser
Compared to Jolicloud, Chrome is in a less developed state. In fact, when you log in, one of the first things you are likely to notice is a statement across the top: "UI [User Interface] under development. Designs are subject to change," with a link to the User Experience section of the Chromium Project. However, after looking at the theoretical concepts behind the design, I would be surprised to see any radical change of direction in the design.
At any rate, to date the main aim of Chrome appears to be the replacement of the desktop with the browser -- a reversal of the traditional relationship in which, because the browser came later, it remains only partially integrated into the desktop.
When you log into Chrome, you are presented with a tabbed browser. On the first tab is a set of icons for online applications. Google's own applications, including Gmail, YouTube, Google Docs, and PicasaWeb feature prominently, although other company's applications are also included. At this stage, the collection is a mishmash of productivity and social networking tools, with a few obvious omissions -- notably, a lack of music sites and faded but still popular sites like MySpace.
Clicking an icon on the first tab opens the selected application in another tab. You can also open another window, which stacks on top of the initial one. This arrangement has the advantage of running all tabs at full size, a potential benefit in a netbook, where screen space is limited, but one that feels restricted and clumsy. Multiple windows can be a pain to manage, and, for all the ingenious tricks, no designer has really found a way to do so. But, working with Chrome's tabs, I soon found myself longing to use them.
I particularly missed adjustable windows when I wanted to compare items -- flipping between tabs is nowhere near as convenient as placing two windows side by side. The only way that the reliance on tabs seems justified is if you assume that nobody will do any serious work on netbooks, and that seems a false premise, considering how popular netbooks are with business travelers.
So far, Chrome includes controls to turn Wifi and Ethernet connections on and off, but no other system settings or customization options, to say nothing of desktop utilities of the sort found in GNOME's applets or KDE's widgets. No doubt some of these features will find their way into Chrome before the final release, but, for now, these lacks add to the impression that using Chrome means giving up much of the control that I'm accustomed to having over my computer interface. While Chrome is easy enough to use, it seems to insist on users doing things its way.
This impression is heightened by the omnipresence of Google applications. You can, of course, bookmark Thinkfree, Zoho and use them instead of Google Docs, or Flickr instead of PicasaWeb, but, currently, at least, Chrome steers you towards using Google's applications.
In fact, it was only while using Chrome that I appreciated how integrated an approach to computing it represents. If it succeeds, Chrome will dominate all aspects of users' computing in a way that no other company except Microsoft has ever done. Admittedly, while Google has a mixed reputation in open source, it still has a better reputation than Microsoft, but I suspect that this control of user experience will produce twinges of uneasiness in open source circles. Even if Chrome is released as free software, the control of computing by a single corporation, no matter how enlightened, just doesn't sit well.
Jolicloud: Once More into the Cloud
The same can be said of Jolicloud, although to a lesser extent.
Jolicloud is an adoption of Easy Peasy Linux, the netbook interface for Ubuntu. This pedigree makes Jolicloud less of an innovation than Chrome, because it still has a strong desktop orientation. The interface is not the GNOME desktop familiar to mainstream Ubuntu users, but rather a top-level menu on the left, and a collection of home directory locations on the right, with the current menu's items listed in the middle. These menus include not only preferences and administration tools, but also standard desktop applications like the F-Spot Photo Manager.
All the same, as I explored the desktop, I was aware of a lack of the usual open source productivity apps, and soon felt -- much as I did with Chrome -- was that the design assumption was that Jolicloud would not be used for serious work. I mean, a Linux distribution without a sound editor, let along one for graphics? One that only includes a dictionary in the Office menu, and no word processor or spreadsheet?
The reason, of course, is that the emphasis is online. You can see that from the fact that the Internet menu is the only one that has a complete set of tools when you first start Jolicloud.
Moreover, as you log into Jolicloud, one of the first icons you encounter is Getting Started. Clicking this icon, you are presented with the chance to register and create your personal dashboard. This dashboard is a combination of a center for your participation in the Jolicloud social networking services, and a graphical package manager called the App Directory.
The App Directory includes desktop productivity tools like OpenOffice.org, but its emphasis is on online applications. Its selection is somewhat broader than Chrome's, and, so far, Jolicloud is not promoting its own selection of applications, the way that Google does. Yet, all the same, I wonder why this administrative option is moved off the desktop and into the cloud. Simply because it can be?
Considering that the dashboard immediately starts broadcasting messages that you have joined Jolicloud and are using a particular brand of machine and are now following Jolicloud, the obvious answer is that Jolicloud wants to keep track of your computing. Although there is no reason to suspect Jolicloud of nefarious purposes, this practice still seems a breach of privacy -- all the more so because the information collection starts without your approval and you cannot stop it without foregoing the use of the dashboard altogether.
Jolicloud is not as large or as dominant as Google in its field. Yet, in the end, the purpose of moving the core of computing into the cloud feels to me like a reduction of my personal control. You don't have to be paranoid to feel that such a move runs counter to the mainstream of FOSS, as well as of basic security practices.
The Wrong Solution
Cloud solutions like Chrome and Jolicloud are efforts to end the division between the browser and the desktop that has existed since the Internet started being widely used. This division is awkward, and, in one sense, I can only applaud efforts to overcome it.
However, after using these two cloud-oriented operating systems, I am more convinced than ever that their solution is not the right one. Going online for social interaction is one thing, but going online for productivity is clumsy, slow, and less private than using desktop open source applications. Nor are online productivity apps remotely comparable to their desktop equivalents in terms of features, even after several years of development.
More importantly, it funnels your computing through your Internet connection, as well as through organizations who can offer no real guarantee that the information that you store with them is secure and will remain uncracked.
Even if you have no reason to distrust the organization (and I have no reason to condemn either Google or Jolicloud as avatars of evil), the fact that the information of countless users is centralized in their services means that, if a security breach occurs, it is likely to affect more users than if everyone kept their information on their own computers, no matter how insecure those computers are.
Naturally, I would not care to see the end of all online applications. But, rather than exchanging a one-sided emphasis on the desktop for an equally one-sided emphasis on the Internet, I would prefer online services dispersed over the desktop, the way that KDE is arranging things, with the option to turn them on or off as local administrators choose.
That, to me seems the way to balance the desktop and Internet connections. Cloud-based solutions only replace one imbalance with another -- and create a whole new set of problems as they do so.