Steven Blank and Start-ups

I heard an interview recently with Steven Blank, the renowned entrepreneur/teacher at Stanford on start-up companies. He is a Consulting Professor at Stanford in the Graduate School of Engineering STVP Program and has been a founder or participant in eight Silicon Valley startups since 1978. His last company, E.piphany, started in his living room. His other startups include two semiconductor companies (Zilog and MIPS Computers), a workstation company (Convergent Technologies), a supercomputer firm (Ardent), a computer peripheral supplier (SuperMac), a military intelligence systems supplier (ESL) and a video game company (Rocket Science Games). Here is a summary of his opinions. What is most interesting is the distinction he makes between large and small. There is no homogeneity between small and large. Small is not a downsized version of large, or a fraction of large. Small and large companies are different entities, different species. And for large companies to create the small company, they must not just downsize, they must mutate.

Blank:
In the start-up building there are no rules, only opinions.
Start-ups are not small versions of large businesses, therefore, there cannot be a business plan.
Business plans explode with the first customer contact.
Start-ups evolve with rapid and nimble experimentation.
The notion of start-ups is really from engineering, not business schools. Business schools are taught by consultants to large businesses. The start-up entrepreneur must escape them, not adapt them.
Large companies are good at innovation only to sustain themselves; it is an elaboration of or improvement on existing products. It is very difficult for them to innovate disruptively. Large companies that have done so successfully have broken their innovative team into small groups and have not allowed them to report to middle management.
Marketing is extremely difficult for start-ups and small companies. Unlike mature companies where the customer knows what he wants and can guide the company, the new market can not be predicted. (e.g. the small dot com companies trying to market on the Super Bowl.)
His three recommended start-up books: Business model Generation by Alex Osterwalder, Lean Start-Up by Eric Reese, and 4 Steps to Epiphany by Steven Blank (Himself)