When Goldacre talks about placebo he tends to talk about the pricing differences between differently packaged but identical medication. Thus, a brand name paracetamol is the same as an own-brand paracetamol and they're the same as a budget paracetamol. Yet one can be five or six times more expensive than the other. (Personally I don't see this as particularly bad behaviour.) I agree that he tends to place too much emphasis on placebo, but then I haven't read the studies. There are strong psychological aspects to pain; pain in cancer patients can be helped with cognitive behaviour therapy; some people (40% in one (not very good) report) with long term pain reported relief when they took a placebo.
But also look at coproxamol. This is an old med. we used to think that mixing meds meant you could use smaller quanties of active ingredients. Coproxamol is a sub-therapeutic dose of paracetamol and a dangerous toxic opiod (opiate?) dextropropoxyphene. (Paracetamol in acute overdose is very dangerous so for the dextropropoxyphene to be the more worrying ingredient tells us something). We've known since at least the 1980s that this med is not particularly effective and is very risky, yet it has only just been withdrawn from general use in the UK. We knew that it was less effective than paracetamol alone (because you'd take that at an effective dose) but that it was far more dangerous and that evidence got stronger through the 1990s and 2000s. (An overdose of paracetamol is treatable if caught in time, although paracetamol overdose does kill many people each year. An overdose of coproxamol is usually fatal before help arrives. For a time coproxamol was the leading medication in fatal overdoses.) Removal of this med was fiercely fought for years by pharmaceutical companies and by patient advocacy groups (sometimes funded by those same companies).
That is sleazy behaviour.
Just defining all effects of pain medication as placebo just isn't scientific at all. Placebos often don't work reliably and repeatable either...