I don't like Software Architects.

What is implicit in the SA role is that the rest of the dev team is full of dummies that can't think strategically or architect big parts of the system.

This naturally leads to the SA producing ever higher and more complicated abstractions that the simian developer tries to implement, poorly, and the SA becomes an irreplaceable high priest -- inflating their salaries and prominence and depressing the salaries of the rest of the team.

Where I work, every developer is expected to be "architect-quality". Sure, more junior folks don't have the experience, but they're expected to have the capacity for strategic, high-level thinking. As they mature, they take on more and more architectural decisions. If they can't do that, then we've hired the wrong person.

Ultimately, that's the only way I know how to scale dev teams and produce amazing software. The SA/Dev stratification is an anti-pattern.

The "writes code" part is key to actually using software architects productively.

The times I have seen architects be productive additions it was essentially a job title for "a super-senior-developer who interacts with everyone." The times I have seen an architect be unproductive or a hinderance it was a job title for "someone who spends all their time telling other people how to code." I've even seen architects who were never coders: since I believe high-level design decisions are intimately linked to low-level design decisions the idea baffles me.

I don't like software architects.

I'd much rather have a team full of A-type engineers that will stop at nothing to force their individual vision into a project.

We can trust that every top-notch developer has the whole project's welfare at heart, and not just the part that they own and whose success impacts their performance evaluation. Ergo, a totally flat team of aggressive rock-star engineers is destined to build their constituent parts in a homogeneous and consistent fashion.

A team of engineering ninjas are bound by a higher power to perfectly coordinate their coding style, core patterns, API design, conventions vs configuration trade offs, third-party dependencies, feature preference, etc., in a way that resonates precisely with the needs of the business and the risks of the industry.

In short, all world-class engineers are business-savvy, have no conflicting interests, never have unresolved disagreements, and are also telepathic. And since all the engineers I work with are obviously world-class engineers, having a single point of accountability and a single vision of quality around is an insulting slap in the face!

I've worked in places where "architect" was nothing more than a title given to senior developers to soothe their egos so they didn't notice they were underpaid. I've all seen architects who didn't code, preferring their ivory towers of UML diagrams. Neither of these types are effective.

A real software architect is a force multiplier: you help teams be more productive through your contributions, be it design, code, mentoring or leadership. Rather than being the team showboat, a good architect is the guy that helps everyone else be even better at their jobs.

I see lots of vitriol here. I think there are a lot of bad experiences, probably I imagine from the ivory tower architects that you get from big consultancies and the like.

I am a proper solution architect. I'm there to serve others with my extensive experience, nothing more. I am always there in the shadows to keep the cogs oiled, provide canned and well tested ideas, stop things that shouldn't happen, start things that should, make sure communication is made and most importantly spend time teaching and helping people. I'm also there to do the risky horrible jobs that no one else wants to be responsible for.

You're supposed to be more a spiritual guide than a facist dictator.

I will say that there is no place for me in an organisation with less than 10 hands on engineering staff.

Reading throught the comments, it's amazing to see how much damage the cowboy culture of our community has produced. "Every engineer should be an architect." Ha, ha, ha...

This also goes hand-in-hand with "Everyone can be a programmer" mantra that everyone likes to repeat. Go teach pointers to 20-year olds and you'll realize how most people can't be programmers.

I don't understand where this culture is coming from. I definitely can't fix my car; when I see my mechanic, he doesn't tell me how I should go learn how to fix my car. I'm confident he's much better at fixing my car than I am. The same is true with other professions -- lawyers don't tell clients to go interpret laws on their own and dentists don't tell people to go remove teeth on their own. What is it about our community that we ingrained into our youth that "writing code" is the thing that everybody must do?

I don't get the hate for software architects. Would you try to build a skyscraper without an architect? No? Then why is having one a problem for a large software system?

Sure, if you're building something small, one person can be architect and general contractor and electrician and painter, etc. But the idea of division of labor is usually seen as a good thing. Not sure why we hackers should see it any differently. If you're building enterprise software systems that span multiple divisions and locations of an enterprise with 30,000+ people in 100 countries, ya know... you probably do need somebody who's job is to look at the big picture, see that the pieces being assembled fit the needs of the firm, and that the pieces work together in a clean and elegant way.

Having an architect, where needed, doesn't denigrate or devalue the role of the developers at all. It really is a different role that needs to be filled.

Anybody fairly handy can build a garden shed. If someone showed up calling themselves an "garden shed architect" and stood off to the side directing traffic I'm sure the "builders" would find it and them to be a complete waste of space. (wait cheesy pun alert?)

However, how many of us would like to get into an elevator ready to shot us to the 100th floor knowing that there was no architect on the project as all the builders decided that was an unnecessary title and those functions could easily be handled by the GC?

Amazon, Heroku etc has made it much easier to build garden sheds - there's a lot more garden sheds around per large building than there used to be. So I suspect if you believe the SA title to be completely unnecessary it's most likely because you're building sheds not buildings. Doesn't mean we still don't want and need them for those larger buildings.

I think of the role of Software Architect as almost a code accountant, keeping technical debt in check.

This sums it up nicely: http://dilbert.com/dyn/str_strip/000000000/00000000/0000000/...

The definition of a software architect varies. In general, mainly refers to those that are responsible for the higher level design of systems.

In my opinion, a great system architect is a great developer, who also have an artistic flair. Those people that, when facing difficult design problems, can often come up with brilliant abstractions. That are so clever, that feel like art. Not only in development, but also in related aspects, like UI.

Those great abstractions will increase the productivity of everyone using the system, and in the end can accumulate to very high amounts of value.

I met Grady Booch at the kickoff meeting for the Society of Software Architects (or some such). OOPSLA 1998 in Vancouver.

I asked Grady Booch "What is software architecture?"

He answered "Software architecture is what software architects 'do'."

At that point I stopped caring.

Until I found the book Design Rules: The Power of Modularity. http://www.amazon.com/Design-Rules-Vol-Power-Modularity/dp/0...

It is the sole source I've ever encountered that had anything useful, actionable, insightful, informative, rigorous, etc.

Alas, I've never been able to synthesis Design Rules' methodology into any of my efforts.

Because what I do is software craftsmanship. I've designed some awesome stuff (and a lot of crap). But nothing rigorous, repeatable, explainable.

For a few years, I bought every software design book I could find. Some of them actually good. But the ones claiming to be about "software architecture" are really describing software craftsmanship. Describe as in descriptive, vs prescriptive.

From memory, Design Rules states that architecture is the set of visible design choices in a product. The entire thesis of the book, backed by oodles of case studies and data, is that deciding where the lines between subsystems, the interfaces, and the allowable parameters for those interfaces, is architecture.

PS- Just read the OP. Nothing actionable. Move along.

The name plate on my desk says "Software Architect", and honestly I feel kind of silly repeating that when people ask what I do. I usually just tell them 'I write software'. That said, I've worked in a bunch of different C#/.Net/SQL Server shops (IT depts, educational, shrink wrap apps, advertising), and I can see in some of these places why the role can be necessary. Not because .Net demands these over-architected enterprise uber-solutions, but because a large amount of developers in those places just don't seem to care about what technology choices and design methodologies are out there. And there's nothing wrong with that, so long as there are some there who are thinking beyond the next sprint, regardless of what their job title is officially.

I'm also pretty sure the role itself means different things depending on the scale of what the company is doing. At my current job (a recruitment advertising company w/ some very large clients), we have a lot of traffic and only moderate development resources in-house, so you really need to depend on proven technologies to prevent the whole thing from falling apart. So part of my job is making sure we are using the right technology where we need to and only rolling our own when it makes sense.

Agree with everything he says. Except for the part that I don't.

A software architect creates a vocabulary to enable efficient communication across an entire company.

That needs to be done with caution, because in a small team/organisation it's absolutely correct. In a large organisation it's just a non-confrontational way of saying "We need an enterprise architecture." That ends up becoming an IT dictionary whose value is to IT alone [1]. And IT serves the business. Not the other way 'round.

If you're doing this in a large organisation, do it for a project, and not for the organisation. It's right up there with the architect that sees an enterprise service bus as a silver bullet. Aka a unicorn.

Edit: [1] Unless you're HP Services, IBM, ATOS, Cap Gemini or some similar systems integrator, because if you're one of them you're in the business of selling billable hours, and an enterprise architecture is an awesome way of billing to eternity without even having to deliver business value.

This is a nice ideal to have. Unfortunately I have never had the pleasure of working with the architect the OP described. I would say the ones I've dealt with have had some combination of: arrogance, stupidity, lack of technical knowledge, lack of creativity, and arrogance (did I say that already).

A software architect job is about interfaces between modules. An architect job is to design interfaces that are aligned with the product direction and technical capabilities (both of the product and the people developing it) and tell a consistent story, both syntactically as well as semantically. Without such a person, each pair of developers, no matter how good an engineers they are, will have to develop their own language that is bound not to match those that are developed around it. This adds to the burden of each developer as he (or she) has to use APIs developed by different such teams. Consistency is key. Because the architect has to see the big picture to do his job, it makes sense that he's also the look-ahead guy. Because he defines interfaces and modules, it makes sense that he gets to evaluate new technologies - so playing with rabbitMQ really is part of the job description. This kind of job tend to require a working knowledge of the product internal, otherwise APIs can get completely incompatible with capabilities. As such, it attracts good developers that also think "strategically". Because many architect still love the code, you usually see such a person wearing two hats - the classic architect one but also the "lead developer" one - so code reviews, cleanup work and super important (but small) modules tend to be part the architect work, even if not in the job description. Lastly, the architect must have soft skills. The job requires a lot of human interaction, considerably more than a developer job does. It also requires a much closer interaction with "management", the architect understands the aforementioned big picture.

Is the architect "better" than any developer? No. It's a complimentary role. Is this role always needed? No, but as a project gets bigger and start having more modules, interfaces tend to get hairy without a guiding hand. So why do architects make more? Supply and demand. Not everyone has the ability or the will to think "big picture" all the time. It's tiring, it's complex and you tend to think about balances (how will this affect the future? How much risk? How many man-hours? What's the alternative? How to phase development?) instead of "code purity".

Disclaimer: I'm an architect but I also do a lot of "lead developer" parts. And while the definition tends to get fuzzy around the edges (and some interfaces are made ad-hoc, without that guiding hand), I see the best results when I follow the above recipe (which, sadly, means a lot less coding than I actually like to do).

TL;DR: architecture is about interfaces between modules, not content of each module. Usually architects also wear "lead dev" hat because they like it.

A software architect I worked with before thought he was gods gift to programming, every time something worked out it was his work, when it failed he always blamed the 'process'.

You should change the title to 'Being a Good Software Architect'.

s/Architect/Engineer/g

None of these characteristics should be stuffed in an architect role. Every engineer should possess them.

Terrifying and silly to think that anyone cares this much about titles

The same could be written of managers and leaders. They exist to serve those they work with, not the other way around. Our society just doesn't seem to understand this.

"A software architect lives to serve the engineering team -- not the other way around."

I thought they are supposed to serve people who are going to use software.

I liked the article. Forget about whether the word "architect" is used by some as a pretense--the author is suggesting essential qualities of a leader. Teams need direction; this seems to me like good advice for a person to provide direction to other peers effectively.

Is it really bad to gloat when right? When people are right, let them gloat. Why does that matter? As long as they aren't trying to make the people who were wrong feel bad about it.

this smacks of noblesse oblige.

("how to live safely in a science fiction universe", apart from being a pretty good book in its own right, really takes this apart)

I liked the article.

I'd like to apologize with some liberty to the sentiment in this thread who deny someone has value because they can't see it. If I may go out on a limb, I'd like to try and apologize for what I feel is being trivialized, but to me, is non-trivial.

I'm sorry every piece of software looks like CRUD and Reports to SA's. Nothing's special like Pagerank special, no matter how many magical automating unicorn wands of libraries and frameworks and methodologies we try to make it seem magical with. Domain knowledge, and experience, on the other hand, seems generic, but it's understanding how the data works, plays, interacts, and exists in all it's states.

I'm sorry if some developers forget that software only answers one question, "Where is everything at?" in different ways for different users, and it's made to be more than it is. We're not out to trivialize anyone's work, however, this point is a rallying point, not something to feel insecure about on anyone's part.

I'm very sorry for all the failed software projects that are inherited for wanting to use the latest, greatest, untested, unhardened. They have compensated well for letting me teach customers how you shouldn't build a house without a blueprint just because you can build a shed without one. No agile mega global waterfall method will get around having to understand the world people face when they use computers in their day to day life. No one cares what we code in, or how we coded. It's up to us to do well, but most importantly, reach a goal for a wider audience without getting hung up on ourselves. I love the latest and greatest tools, but I know better than to put it into production somewhere to be kind to the future me.

I'm also really sorry to myself that I might have learnt some skills I may never use or enjoy using, like overseeing a full cycle ERP implementation.

It did help tremendously, though in being able to find and connect the dots in intricate relationships between data and be able to put a picture together that makes sense for everyone beyond me.

SA's existing do not mean anyone is a dummy, or smarter. It just is, what it is. Trivializing what anyone does is a pretty close minded, harsh judgement for the intelligence and open-mindedness here. Just because one person can't imagine a position having value, doesn't mean the value can't exist.

Architects get thrown down more wells than they deserve and smart developers are always open to learning from the rattlesnake biting the other person. It goes both ways. Great architects listen, listen, listen. They provide the foresight that today can't always show. Good architects code. Most I know thrive on Proof of concept coding things.

If there's SA's that don't stay current (meaning, they don't keep hacking with new stuff), that's something that can be faulted in anyone.

We can look at it the same way with developers.. having a legion of coders who have not yet had a relationship with a code base for more than a few years don't really learn the reality of being kind to your future self in code, thought, and trivialization of what anyone does.

Maybe some part of this is from having been online for almost 2 decades, and building web based software for over a decade. If I'm off, I apologize and ask for your tolerance.

Our coding isn't special. Frameworks come and go. All glorious languages worshipped today were made in the 90's. Any code that is 2 years old is legacy. So, what is your legacy?

Mine is writing code that can outlast and not need me anymore and can be well structured and maintained without me. Why? So I can go solve the next problem.

Connecting a need to technology and making ourselves invisible is what I really enjoy doing, but maybe it's just me.

Read Cringley's Accidental Millioniares book for lots of background on the "Software Architect" idea, particularly at Microsoft.

Huh.

Where I work, Software Architects are people who get paid lots of money to dick around with "new" technologies like RabbitMQ. Then after spending seemingly weeks implementing a trivial proof-of-concept app with the new thing, they present some slides and Visio diagrams at a brown bag nobody attends, and the rest of us ignore them so we can keep building things that actually provide value. Then they go on vacation to Aruba (probably).