The thing I like about Jason Fried and 37 Signals is their tight focus on what they do. They are at once their own clients, customers and dev team. This gives them a great deal of freedom when it comes to features, functionality and process. However companies like 37 Signals are definitely in the minority, and most of us have to deal with much wider range of issues and stakeholders.
The problem I have with 37 Signals is their style of writing. If you've read their book or any of their posts, the stuff they talk about is rarely presented in the context of their specific working environment. More often it's presented in terms of absolutes. I'm never sure whether this is merely stylistic, a desire to create link bait, or the result of a lack of empathy. In truth it's probably a mix of all three.
Quite often I find that they misappropriate cause and affect. For instance, they had a developer in Europe and their first project was a success, therefore the geographic location of their developer was partly responsible for that success. In reality the guys at 37 Signals are really smart, and would have been a success whatever process, language or geographic set-up they had. However there is a tendency with genius design not to realise that's actually what you're doing, and attribute success to other factors. Sort of like the Olympic medalist who attributes her success to a lucky charm rather than raw talent and years of experience.
A good example of this didactic style can be witnessed in Jason's recent post about Personas. As the article explains, 37 Signals don't need to use personas because they are essentially building products for themselves. However because personas present no value to 37 Signals, they suddenly seem to present no value at all.
Jason argues that personas aren't real people. You can't ask them questions and they can't tell you when they get frustrated or when something is wrong. This is all true, but its also misdirection because this doesn't actually have anything to do with the value of personas.
For a start, personas don't substitute the need to do your homework, talk to real people and test your assumptions on a live audience. In fact, the best personas are created out of exactly this type of research. As Jared Spool rightly points out, you shouldn't mistake crappy personas for personas being crap.
I totally agree that you don't need personas if you're building something for people like yourself. However in an agency environment you don't usually end up building sites for other web designer. In the past year we've build sites for everybody from scientists and teachers to environmental activists and mobile game buffs.
The problem many companies face is that they **think** they are building the site for people like themselves, when in fact they aren't. A 40 year old technical director will have a very different outlook on life than a 16 year old girl. In fact, time and again our field research has shown that client assumptions about their market can differ substantially from the market itself.
So research is definitely important, but how does this feed back into the persona argument? Well, if you're a small company and all your team have immediate access to your user base, maybe it's easy to ring a few people up and ask their opinions at every step of the way. However for larger companies, personas are a really useful tool for summarising and circulating the results of your user research. They are also a great way for framing internal discussions about user requirements. So rather than talking about a homogenous "user", you can talk about "Bob" or "Mary". Personally, I find personas are most helpful during the discovery phase, when you're building up domain experience and empathy with the users.
That being said, personas are just one of a number of tools at our disposal. I don't think any user experience person worth their salt would say that persoans are required on every project. Furthermore, it's very easy to overstate the importance of personas. In fact, this is something we've been guilty of in a few recent projects. The biggest problem with personas is the fact that they often become just another deliverable and end up sitting in a draw unused.
I think the argument about personas is ultimately one of context. It's ludicrous to argue the merits of a screwdriver without knowing the situation. A screwdriver is great if you're extracting screws, but useless if you're trying to undo a bolt! Similarly, if you're building a site for a group of web designers, you probably don't need personas, whereas if you're building a site for a group of doctors, they could come in handy.
But I guess being logical and rational doesn't create the same stir as being sensational, hence the title of this post.