I think Lisp is simply the right choice for great applications, whether entrepreneur driven or otherwise. For Corporations, if the managers don't understand the value of Lisp, they will opt for different solutions, even if those solutions destroy the value that Lisp offered. Perhaps the only barrier to wider Lisp adoption is an education barrier?
I have personally seen Lisp stamped out for political reasons. I had been tasked with generating C code from a textual description of a finite state machine. With my boss's permission (also a Lisp programmer) I used scheme to take in the descriptions and output C code. It hooked right into the Makefiles and worked well. Despite the fact that the solution worked and that there were several engineers on staff who were Lisp programmers this did not make it through review. A senior engineer who had been with the company for a long time (who'd never programmed in Lisp) had major objections. He pulled the right strings and we were forced to reimplement it all in C. I suspect one of real problems for Lisp is the presence of this sort of anti-Lisp person in the industry.
If you had to reimplement in C then it suggests that whatever non-C language the original was in, e.g. Python, you would have had to re-write it in C. Otherwise, you would have had the choice to reimplement your Lisp in a more suitable language than C that was acceptable. Lisp here seems a side issue.
You are probably correct. There were snide remarks made about Lisp in particular however, like claiming it is not "used in industry" and is thus "nonstandard". Similar remarks could have been made against any language the guy DIDN'T know. Regardless of the reasoning there are people like this who end up stamping Lisp (and other non-dominant solutions) out.
Yes, it's an education barrier. BRL (Scheme adapted for web apps) has survived numerous attempts to quash it at my employer over the past 7 years, as different managers have needed educating.
The other education barrier is with programmers. The "Wow, this is simple and useful!" response comes in 15 minutes using Python or Ruby. With Scheme or CL it comes in a week or two, though it's stronger. I don't know if s-expr notation will ever have immediate appeal, or if it needs to grow on everybody.
Taking two examples of applications written in Lisp and trying to shoehorn a premise to it is ridiculous. Lisp/PHP/Perl/Ruby/Java are all tools. Anyone who tells you that one house is better than another because it was built with a different type of hammer would (and should) be laughed at.
Take this quote and substitute Ruby/Rails or Python/Django or even Java/Spring would give you the same answer and even have a larger base of developers to draw from.
"Our hypothesis was that if we wrote our software in Lisp, we'd be able to get features done faster than our competitors, and also to do things in our software that they couldn't do. And because Lisp was so high-level, we wouldn't need a big development team, so our costs would be lower. If this were so, we could offer a better product for less money, and still make a profit. We would end up getting all the users, and our competitors would get none, and eventually go out of business."
Are you saying that if someone built a house with an inflatable mallet and someone else built one with a nail gun there would be no difference of quality in the resulting house, or that one wouldn't be completed sooner, or be more extensible?
(Yes, you can laugh at me.)
I agree with you that all languages are tools (I love working with all kinds, actually). But aren't some tools better at some things than others? Or what would we need so many for?
I'm not saying Lisp is the nail gun and everything else is an inflatable mallet, but I do think there is a difference.
What I got out of the article is that Lisp is a tool that worked pretty darn well in a couple cases. I feel that's a good thing, because I kind of like Lisp.
I'm saying whats important is the quality of the resulting house (we maybe extending the metaphor too far here). If you can make a better/higher value/more usable house with an inflatable mallet then with a nail gun then yes that may be the right choice.
"But aren't some tools better at some things than others?"
Absolutely, and note I'm not taking shot at Lisp as a language. My argument is with the article, there is nothing here that supports his premise that Lisp is for entrepreneurs. That it was the right choice for a couple of cases? Sure! but for anyone starting a software business? nonsense.
Languages are not just hammers. They are universes for thinking in.
So you're using Rails... What if I told you all your competitors were using Java and Spring? Gives you a surge of confidence, doesn't it? That's the feeling Lispers have about Rails programmers, for the same reason.
Very cool post. I don't think this is limited to Lisp though. I bet that if a big company acquired a company whose product was written in Python/Ruby/etc then they'd consider rewriting into C# or Java. And they'd have the same issues that Yahoo and Sony had, as expressed in this article.
With all due respect to the Sony and Yahoo engineers/managers I wonder if the shortcomings of the reimplemenations have less to do with the choice of programming language and more to do with the characterestics of people doing the work. I have a suspicion that the "Lisp startups" of the past tended to be composed of particularly intelligent people.
IIRC their main issues were threading instability in CMUCL on BSD or mac, which I believe has since been fixed in SBCL. In general, SBCL has come a long way, partly due to companies that use CL (e.g. ITA) funding its development. CL is in many ways much more production-ready than it used to be for that reason. It still has rough spots, but my experience is vastly improved over what it was 1.5 years ago.
I'd say, XP (extreme programming) is for Entrepreneurs.
Not only for coders, for everyone that needs speed and quality that are variables on opposite sides of the equation.
