A sleazy way to avoid direct PCR Sequencing

[Seth Findley]

Everybody knows that sequencing a single subclone from a PCR reaction 
isn't too smart because of Taq-mediated mutagenesis.

And direct sequencing of PCR products can be a technical challenge...

Can anybody tell me what might be wrong with the following technique 
which I used recently to sequence something I PCR'd?

In order  to increase the complexity of the template I wanted to 
sequence, thereby avoiding individual artefacts I did this:

After cleaning up my PCR reaction and polishing the ends (I have a very 
efficient protocol for this) I subcloned the fragment into a very well 
phosphatased (blunt) vector and transformed some E. coli.

Now, rather than pick ONE colony for sequencing, I scraped about one 
*thousand* colonies off the plate and resuspended and mixed (well) the mass 
in a small volume of medium. Then I used a bit to inoculate an overnight 
with. I did an Acid-Phenol miniprep for sequencing and got nice clean 
sequence.  (I used an internal primer, so any vector subclones wouldn't 
be primed by my internal primer).

In this way, I was able to sequence many different PCR products with 
little headache.

Now I know this looks pretty retrograde, but for somebody who might not 
be able (for whatever reason) to sequence PCR products directly:

Is a thousand colonies enough to be pretty sure of the sequence?