That decision was acceptable because of the complete lack of any technical experience in house (100% outsourced, multiple vendors) who had very different patterns of development. Couple this with no tests, no process and no patterns, you're stuck trying to work against something that's unpredictable.

The way forward then is to hire a team (even if it's just one guy) internally who can have ownership of code. Then instituting good practice and a sensible method of rebuild and rediscovery. Obviously this hinges on having enough runway to get into it properly.

tl;dr: sometimes it's ok to scrap it all and start again when your goal isn't about preserving code but instead is about building a better product and technical culture inside a company.

Getting downvotes - I wanted to add something: There's a real risk of making problems worse when you don't have any ability to stop and change. Working with existing code is great but when you have existing customers running into bugs on a daily basis, and fixing them would change the business logic that others are relying on, you either choose to build more technical debt or you swallow it and start again. Obviously this depends on runway, position, etc.

There are many times when companies go through their proto and raise seed only to find that their original hypotheses aren't working, and to change requires a significant re-architecture. Bolting this on to existing codebases can end up with code so complex that it won't stay up, and you end up spending six months in pure fire-fighting ops mode to keep the thing alive. Knowing when to recognize this and acting accordingly is the right approach, even if that sometimes means calling it a day on the existing code and starting again.

You have to keep the train moving. That keeps the cash coming in. You can't stop the train.

So, you have to take what you have and somehow -- that's what you are being paid for and hopefully you have the chops -- manoeuvre the beast into a better state. This is fscking hard, especially when you have to get the darn thing under test. And it's invariably massively undervalued by most biz management teams.

You can _always_ extract value from operative but shitty code, even if you have to stick your head around the U-bend and pull it back with your teeth. If you don't know that yet, then you've more to learn.

Also, when the real value within the app is a simple product, breaking it out and re-engineering based on what you've learnt takes about roughly the same time as it does to walk through the code line-by-line and add in valuable tests.

Refactoring legacy code by multiple vendors is always a risky proposition (and much much harder to sell to clients who want a deadline for delivery).

If you give in to them too many times and end up with buggy code where trivial changes take a month, but has all the features that are needed where do you think the blame will fall? I can guarantee you it isn't going to be the business side.

The goal shouldn't be for the CTO to be the business side's technical henchman. It should be to look after the technical side of things. If a rewrite is the best decision technically (without considering the business), then that's what the CTO should advocate, but they need to be willing to work with the business side to find a solution that works for both sides.

Engineers are the only ones who can advocate for pristine code without tradeoffs, and I would argue that great engineers care enough about the business to not do that. Really great engineers just build the pristine code in the same amount of time regular engineers would have taken to get a simple version almost-working.

I actually like the idea of "technical henchman", though!

This has been key in my experience. I'll build a feature that is flexible, clean, well factored, and completely documented in the same time it requires someone else to make a half working mess.

CTO's as I'd like to understand them bring the bar as high as they can within business constraints. Doing any less is accepting mediocrity. Is that something you'd want in an executive leadership role?

Look at Windows Phone 7.

All kidding aside, that is the central question of being a start-up CTO. All non-business building aesthetics and measurements are out the window until you have customer #1. After that, the goal is to support the customer as best as you can. If ugly code is making modifications difficult, then you rewrite only then. If it is working, then for all that is holy leave it alone. If the code is beautiful and a change will make it ugly, then it will be ugly.

One of the lessons I wrote about goes core to this. There is more value (to a company) in learning how to navigate, manipulate, and fix large bodies of ugly/shit code than trying to rewrite it in a perfect way.

Good CTOs know about all the shiny new things but have the good taste to apply them only when appropriate, can manage the finances of an engineering team, and aren't afraid to hire developers who might be smarter and better than themselves.

Bottom line is that you base your strategy on what the company needs, not on some abstract notion of a good CTO. The only consistent objective is to make yourself as superfluous as possible by creating a balance between engineering quality and business objectives that maintains itself without constant management interference.

I realize the audience here on HN - but that's a very important distinction to make.

The world is full of tons and tons of successful, large-scale software systems (from apps to languages to databases to OSs) that were not only "rewrites" of older systems, but had to be written because the original system had incurred too much technical debt.

So what Joel really meant, deep down inside, was "okay it's not that you should never replace large systems from scratch; it's just that you should be very circumspect about it, OK? You don't just go knee-jerkedly ripping out the old stuff just because it's a ugly, a pain to deal with, unsexy, etc. You have to look at the whole picture, and understand the (true) business costs of what you're doing. That's all I'm saying."

What I took away from the article is this: whenever you're thinking of rewriting from scratch, there's a better way - incrementally replacing the code. It's not as fun, and it feels worse, but it ends up with much better code. I'm pretty sure this is nearly always true.

Joel outlines the reasons, but the most important one, imo, is this: you're not going to do a better job. You think you are, because you see that the code is a mess and think you're better than the people who designed it. And you're probably right (or at least, you know more.) But a system that has years of work on it is not something you can whip up again from scratch - you're going to miss all the same things they missed the first time, when they were designing it, and either recreate all the bugs, or make your new system a mess.

Incremental improving is nearly always the saner solution.

(Or was this just a joke and I missed the point?)

Anyhow, the decision was quite clear a rewrite was in order. The original code base received a little love here and there to make it work, but in the end we replaced it all and the company ended up selling for a little shy of 8 digits.

While you may not be able to just scrap it all, there is no reason you can't do things the right way behind the scenes and then relaunch the right way over time. I think a good CTO or leader should recognize what will be better for the company, and perhaps at times that does indeed involve maintaining what you have - but if it is indeed broke; it's also in your best interest (as well as the company's) to fix it.

"There are two classes of Free Electrons. Sr. Electrons and Jr. Electrons. Both have similarly productivity yields, but the Senior versions have become politically aware. In technology savvy organizations, most CTOs fall into this category."

http://www.randsinrepose.com/archives/2005/03/20/free_electr...

It's easy to think about what the perfect greenfield solution would be, but this is rarely the case. There's the "best" solution, and the "most applicable" solution.

In my conversations with non-technical managers I spend much more time arguing against the "best" solution, and for what best fits our situation. If I can't reach them I will break out this line:

"Nothing would make me happier than to spend more money and time on new technology, but I think we can meet your needs with X."

It's not about whether starting from scratch is okay or not. It's not about whether mixing various technologies to get to something that works is okay or not.

It's about aligning business needs with the execution plan and the resources available.