Google Web Accelerator, released yesterday, is a free browser plug-in that speeds up web surfing by routing all web page requests through Google servers and by caching web pages in desktops before they have even been requested.
The software is designed to make web browsing faster, using an interesting combination of a few old tricks involving client-side and server-side caching.
When GWA is installed it requests web content from Google’s data centers, rather than the origin site, potentially speeding up the download. The content will be cached at the data center and on the desktop, in cache separate from the browser cache.
There is also a pre-fetching function. When a page is requested, the GWA toolbar will also pull certain linked pages down into the local cache, so they can be accessed more or less instantly if those links are clicked.
It was not immediately clear how Google determines which pages to pre-fetch, and the company did not return calls for comment by press time. There’s an obvious denial-of-service risk, unless pre-fetched pages are selected carefully.
The company does tell webmasters that they can specify which popular pages are pre-fetched by the GWA toolbar, by adding some metadata to their HTML links.
While it can be turned off, some of Google’s new rivals think that this pre-fetching function means the service is unsuitable for dialup users with low bandwidth connections.
They won’t compete with us directly, their focus is primarily on broadband, said Dave Fannin, CEO of Propel Software, which sells software very similar to GWA to ISPs with large dialup subscriber bases, and has about 35 million users.
Google has not said how or if it plans to monetize GWA. It could decide to sell to ISPs, following the Propel model that it software already mimics. Or it could trying to use GWA as a way to optimize advertising delivery.
The caching concepts are not new to the internet, but Google’s brand and user base behind it could prompt an unprecedented uptake of this type of web acceleration technology, and that has privacy watchdogs worried.
Google already has products that know what you search for on the web and what documents you have on your computer. GWA means it will know what its users surf for, even when they do not use search to get there.
But it’s not well known what happens to that data. It’s broadly thought that the company just stores everything that passes through its servers, forever.
Google’s data retention policies are anything but transparent, said privacy advocate Lauren Weinstein, co-founder of People For Internet Responsibility (PIFR).
Google’s privacy policy says that the data that passes through its network is pretty much the same as the stuff that passes through a regular ISP’s network, IP addresses, HTTP headers and the like. There’s no word on storage.
Privacy advocates such as Weinstein are concerned not with Google intentions, but with the intentions of a person with a court order or subpoena, or a skilled hacker, who could access this treasure trove of user data.
If Microsoft started doing this, people would be screaming bloody murder, Weinstein said. Microsoft has this Darth Vadar image. Google has this smiley face do-no-evil reputation and they can get away with it… Even if you accept they’re good, the data alone is the problem.
GWA is supposed to work on Internet Explorer 5.5 and above or Firefox 1.0 on Windows 2000 or XP, although, as beta software, online user forums are already tracking apparent compatibility problems.
