How to read a Microsoft white paper

As part of its marketing efforts to promote Vista, Microsoft UK recently released a white paper that highlights the company's claims about improved power efficiency when using the new operating system. In particular, it promises that companies "can save anything from £23 to £46 per system per year" by switching from XP.

The company came to this conclusion by paying a lab to put a selection of desktop and notebook PCs through some tests, and to do a straw poll of PC users about their habits. Microsoft is up front about the fact that Vista uses no less power than XP when the computers are being used — instead, it says, the new operating system is far more sensible about shutting things down when not in use. The company further claims that the addition of power management to the group policy system makes it easier for corporates to enforce power-saving configurations.

So far, so good. The first three sections of the paper describe what the various power-saving configurations, such as sleep and hibernation states, are, and why XP isn't very good at using them. When we get down to the figures, though, some curious choices have been made. The headline figures are there — "dramatic savings, especially for the desktop PCs" — and for good effect, are repeated in three tables; once by themselves, once multiplied by 200, and once again with an additional figure for further savings if you use the group policy management to fine-tune things. That makes a difference of just a couple of pounds a year, which Microsoft bigs up by multiplying it by a thousand in case you're running that many PCs. But there's really only one comparison — the out-of-the-box behaviour of XP versus Vista. Later on, the same data return in yet another table, this time multiplied by the standard conversion factor that turns energy used into kg of carbon dioxide.

What happens if you bother to enable the power-saving features of XP? Microsoft doesn't say. Rather, it says that people don't do that because the controls are hidden away, and that because third-party drivers sometimes override the system, you can't get a reliable sleep state. Another way of looking at that is to say that XP is hard to manage and not reliable, which may come as a surprise to companies that have been managing XP perfectly well for five years. Assuming that you do have an IT manager who can in fact manage IT, then what figures would we be looking at? That wasn't tested — or if it was, the results weren't fit for human consumption.

There's more fun and games. For reasons never fully explained, Microsoft publishes power consumption figures for the laptops and one of the desktops under Vista running Aero, and with Aero disabled. The figures for the two modes are almost identical, certainly within the range of measurement error, and that's as you might expect. This is the only time that actual wattages are used instead of cash or carbon dioxide equivalents, which makes any sort of comparison with the other figures difficult.

But then, that's all we get from the notebook tests. How notebooks work under Vista instead of XP is never mentioned: Microsoft says that this is because laptops work differently and use power to charge the battery even when in sleep or hibernation. Undoubtedly true, so why test them? The phrase "especially for the desktop PCs" mentioned before certainly implies that there was data for the notebooks at some point. The way all the tables actually published keep repeating the same data in different ways also hints at some last-minute editing to replace figures that didn't make the final cut. Microsoft said that it didn't want to produce a report "500 pages long" and that most people use desktop PCs anyway.

However, once you start reading the text with the idea that the notebook data was there and subsequently removed, there are plenty of hints. For example, explanations start "For desktop PCs...", but never get around to saying how notebooks compare. We can surmise that the figures weren't very good, and that the basic single important fact behind the white paper — that people don't enable power saving on XP — isn't actually true for notebooks.

The paper concludes with some useful general observations about saving power and some pointers to other places where these issues are discussed.

It's certainly good that energy issues are being talked about, and that Vista has sensible defaults. More than that, the paper reads like an exercise in expanding a single valid point well past its applicability, and without mentioning other implications and alternatives. Sky TV has just sent an over-the-air update to its Sky Plus digital video recorders, for example, telling them to go to sleep after a period of inactivity. Could such an enabling update be sent to XP? And if it could be sent, would it be sent — given that it would nullify a commercial advantage for Vista?

By showing that Vista behaves so similarly to XP in most uses — and not discussing modes where it provides possibilities that XP just can't match under any circumstances — the white paper brings to mind the one major consideration for Vista upgrades that Microsoft likes least. Why bother?