A brainless comparison of JPG and RAW that apparently primarily relies on the author having no eyes in their head. JPG always has artifacts, unless it's saved as uncompressed JPG. But uncompressed JPGs are significantly larger than RAW, since RAW simply stores the sensor data (one color value per pixel!) instead of three color values.
Furthermore, he conveniently overlooked the fact that JPG only offers 8 bits per color channel, whereas RAW typically offers 10 or 12 bits per color channel - which preserves significantly more detail and offers higher dynamic range. Plus, you can then make far more adjustments to this data that don't destructively affect the information in the image.
But of course he only makes perfect images on the first shot and therefore doesn't need any later post-processing. However, the fact that even simple shrinking or enlarging of a JPG means decompression of a lossy compressed image combined with equally lossy recompression, with bicubic interpolation for the shrinking in between - he conveniently ignores that too.
Anyone who uses JPG directly in the camera instead of RAW has a fundamental problem: they're essentially throwing away their negatives and keeping only the first print, from which they then make further prints. No professional photographer would be so foolish with analog media, but apparently there are always people who think the basics of value preservation (which is what a professional's business is built on) don't apply to digital technology...
To his criticisms: my DCS 520 does 3.5 images per second. And that with RAW formats - it primarily only does RAW formats. In a burst it can handle up to 12 images. Newer cameras have similar numbers with significantly higher pixel counts (the DCS 520 only has 2 megapixels). JPG and standard TIFF can be produced in the background if needed - while the camera isn't currently photographing, it converts images. This way you have ready-to-send images available and still have the digital negative available.
As for file size: the negatives are under 2 MB, which isn't a problem with today's memory cards. The same applies to other camera systems with larger sensors; RAW formats are typically in the MB range roughly equal to their megapixel count, at most twice their megapixel count.
The inconvenience of conversion is also relative: an installed Photoshop plugin and I can open the negatives directly if needed. Then I can apply Photoshop's normal batch methods to them. Or I can convert the files using Photo Desk software, which only takes a few keystrokes. If he has to wait for conversion before continuing to photograph, that says less about the RAW format than about his complete inability to build simple automation scripts. And the latter is rather embarrassing for a so-called professional - after all, workflow efficiency is supposedly the alpha and omega of a professional, so you'd expect basic engagement with your tools. And as for the image content being the same in RAW and JPG, that guy is just blind. I deal with JPG frequently enough (for example, my image albums on http://leicaesk.de/ dynamically produce JPG from the original PNG files), the artifacts are clearly visible when you look at fine structures. A good example is ((deleted image, sorry, shut down the server it was on in 2007)) - just go to the largest view and look at the horizon where the wind turbines are. The image was captured directly as JPG; my Olympus E-100RS is just too slow when I shoot TIFFs and the camera can't do RAW. The artifacts around the wind turbines are clearly visible as streaks or shadows. With multiple recompression, the effects get larger. In RAW you have clearly defined contours there without the streaks.
Sure, RAW is proprietary format that needs special software to handle. Of course you should back up standard format files alongside your RAW files - for information preservation reasons TIFF is suitable, since you can use 16 bits per color channel there and color profiles find sensible use. The file size isn't unproblematic there, but the alternative would be loss of information you can never reconstruct again (except from the original negative).
What's certainly true is the speed of some RAW solutions: but that's a software problem. If it's too poor, you should simply consider whether this camera combined with that software is simply the wrong tool.
So my conclusion would be different: shoot RAW whenever possible. Choose camera system and software based on how efficiently this workflow is implemented and how meaningfully the features can be used. Test camera and software as a unit - especially if you want to make money with the tool, the whole chain should work. Petteri Sulonen is of the same opinion; he develops this further and gives good hints.