The lesson of this last post is not about gmail in particular; it is that web-based software, provided as a service, isn’t going away. If anything, it will keep expanding, because the user benefits are of a sort that traditional, user-managed software will have an extremely hard time matching.
It isn’t just that web-delivered software can be very flexible (especially once greasemonkey is involved), or as powerful as all but the most powerful rich client software, or that it can be really convenient to have access to your data at any time and any place, or that it is nice to be social and trivially share things with friends. All those things are very nice, but none of them are particularly exclusive to the software as a service, and traditional software either already does better or can catch up if it wants to. If these were the only questions, I’d put my money on locally managed software.
But these relatively shallow software features aren’t the only issues. The problem here for any provider of locally-managed software, be it the Free Software Foundation or Microsoft, is that software as a service is a different architecture; an architecture which provides features which go outside of pure software. Most importantly, this architecture abstracts away the most hated and unreliable parts of the self-managed software ecosystem- hardware, support, security, and maintenance. Those are someone else’s problems- all you have to do is log in and use the software. In Jesse’s words, ‘I no longer have to worry.’
In the locally-managed software world, those issues can be truly resolved only with redundant hardware in redundant locations, reliable bandwidth, complex mirroring setups, and the application of lots of manpower, both to set things up and to minister to them when they go wrong. You can improve every part of the system, but the need for time, maintenance, and redundancy will never be completely eliminated. Hardware and software will always require maintenance, and time and skill will be needed to resolve the inevitable failures. The million dollar question is whose responsibility these things are. Hosted software promises to make that responsibility go away, so that you can focus on other things and sleep easily at night.
In an age where everyone has gigabytes of data to back up, hundreds of pieces of software to keep up to date, and so on, this ability to sleep easily at night – to not worry – to put the responsibility on someone else’s shoulders – is not to be undervalued. People will make many compromises, in functionality and in other freedoms, in order to reduce that worry and get that security. Of course, the security provided by some (all?) of the hosted service providers is to some extent illusory. Hosted service providers can be subpoenaed, or fail, or decide to hold your data for ransom. But people strongly believe (with some reason) that software and hardware are even more likely to fail, and at high cost given the centrality of our data to our lives. So until that expectation changes (either because service providers get worse, or because self-managed software gets radically better) software as a service will only become more common.
The implications of this for personal freedom will be the subject of an upcoming post; suffice it to say right now that we need to start thinking about principled services now so that we can design and implement them.