I've seen this kind of strawman argument against Agile before, though this one builds a bigger and more detailed strawman than I've seen before. The problem here is not Agile, it's slapping the label "Agile" on something and then doing something totally different.

The great thing about Agile is that it's not business driven. I mean, of course you the work for business reasons, and the business people who have determined what they need, need to be available to the Agile team (for example in the form of a Product Owner in the case of Scrum), but in the end, it's the team of developers that makes the decisions about how they do it.

Of all the claims that this article makes about Agile, I haven't spotted even a single one that's correct. All of them are quite obviously false. The article is a piece of junk.

It seems like you are invoking the No true Scotsman fallacy. There's a lot of companies that do "Agile" and "Scrum" that are exactly like he describes. When Feature X has to be done this sprint because it's already been sold to (Board|Client|Investors|Other Dept), there's not much time to think about "How?", it's more like "If I'm going to be working Sat and Sun again?".

I'm sure there are organizations out there where Agile/Scrum is working fantastic for them in practice, and developers are happy with their work and career prospects. I just have yet to encounter one in the wild personally.

In what possible way is that a No True Scotsman fallacy? Do you think you can slap a "Scotsman" label on a fish, and it magically turns into a Scotsman? What Agile is, is described in the Agile Manifesto. What Scrum is, is described in even more detail. When what you're doing is not that, you're not doing that, no matter what you call it.

I totally agree that there are a lot of companies that claim to be doing Agile or Scrum, and often it just means they picked one element, pulled it out of its context, and applied it in a situation that has nothing to do with Agile or Scrum.

Years ago I worked for a company that claimed to be doing Scrum because they started each day with a stand-up meeting. The meeting was fine, but what they did really wasn't Scrum by any stretch of the imagination. No sprints, no backlog, no retrospective, no well-defined stories, and no clue what our pace of development was. And even if you do have all the trappings of Scrum, it's still not Scrum if you ignore the underlying principles.

Words have meanings, or they are useless. So, let's define "agile". It doesn't mean "apply a label to any process that you already have". So, where do we find a definition for agile? Probably the best candidate is the Agile Manifesto. But there's not much concrete process there, so now what?

I worked in an agile (and, in fact, XP) environment for a while. The rule there was that the product manager owned the decision of what stories went into the project, but the technical team owned the estimates. That's kind of agile-y (we're doing stories and estimates, yay for us!), but it's also just good engineering management.

If you're in a situation where "Feature X has to be done this sprint because it's already been sold to (Board|Client|Investors|Other Dept)", you're in trouble whether or not you call your process "agile". The trouble you're in doesn't have much to do with whether you're agile or not, either - it has to do with management that doesn't listen to engineering reality. (However, it is a violation of at least one of the principles from the agile manifesto - sustainable pace.)

> So, where do we find a definition for agile? Probably the best candidate is the Agile Manifesto. But there's not much concrete process there, so now what?

So, now we recognize that "agile" doesn't describe the elements of a software-development process, it describes a set of priorities which are used for determining what process is adopted, and how the processes adopted are applied, in a particular organization (a set of priorities which specifically calls for consideration of the particular people involved and how they particularly interact.)

One of the principles of the Agile Manifesto is "People over process". While that doesn't mean process is not important, it does mean people are more important. Whatever your process is, it should be focused on people, working code, customer collaboration and the ability to respond to changing requirements. Even if you try that, it's still possible to have a process that does these things badly. But I still have to see someone argue that these principles themselves are a bad idea. (Well, responding to change can be harmful when driven to extremes.)

Scrum is the most process-oriented process that's still pretty agile. But one of the most essential aspects of it is that you refine your process based on how it works. In an organization with multiple Scrum teams, eventually every team will have a slightly different way of doing things, because that's what works for them.

Fair enough. But that means that the complaints about agile don't really have much to do with agile, they have to do with bad processes that are trying to implement agile. (Scrum is more concrete, though, and can be more validly criticized or defended without as much confusion over the definition.)

> But that means that the complaints about agile don't really have much to do with agile

Well, yeah, that's usually the problems with complaints about agile -- they aren't about agile, they are about a specific set of processes (usually ones developed, adopted, and applied by an organization not applying the principles in the Agile Manifesto) which that particular organization, or the consultant selling them, called "agile" as a way to hook on to the popularity of the buzzword.

> Scrum is more concrete, though, and can be more validly criticized or defended without as much confusion over the definition.

In principal, sure, Scrum is subject to process-based criticism, although lots of criticism of Scrum are about "what I've seen organization X, that claims to be using Scrum, do" rather than "the framework defined in the Scrum Guide".

> It’s probably not a secret that I dislike the “Agile” fad that has infested programming. One of the worst varieties of it, Scrum, is a nightmare that I’ve seen actually kill companies.

Scrum is not a variety of Agile. Agile is, per the Agile Manifesto, an approach to software development which prioritizes (among other things) individuals and interactions over processes and tools.

Scrum is a tool (a process, arguably, though it describes itself as a "framework" as opposed to a "process", since any complete process will need to add to what is presented in the Scrum Guide) laid out in a "definitive guide" providing the "rules of the game". An organization might be applying Agile principles and use Scrum, but those are largely orthogonal concerns.

> So what are Scrum and “Agile”? I could get into the different kinds of meetings (“retrospective” and “backlog grooming” and “planning”) or the theory, but the fundamental unifying trait is violent transparency, often one-sided. Programmers are, in many cases, expected to provide humiliating visibility into their time and work, meaning that they must play a side game of appearing productive in addition to their actual job duties.

This does describe the actual process in many places, but it has nothing to do with "Agile" or "Scrum". It is a fairly common management failure that was common among the canned one-size-fits-all processes many things that the Agile Manifesto was a reaction against; Scrum -- as defined in the Scrum Guide -- insofar as it incorporates what might be described as "violent transparency", does not do so in a one-sided manner.

"Lots of things that are called 'Agile' have problems because continue to do things -- which are also common without the 'Agile' label -- that the Agile Manifesto was a direct response against" isn't a substantive criticism of Agile, though its a good warning -- as is, after all, the Agile Manifesto itself -- against buying a canned process or tool and applying it just because the person selling it says the word "Agile".

The comments on the main article are too long. This isn't a problem with Agile/Scrum. This is a problem with poor management and not setting up teams for success. The entire point of Agile is to enable teams to deliver on their own schedule and by their own rules, with the team made up of all the people who will enable that to happen.

The OP makes real points about the failings of Agile. The Engineering team ends up subject to the whim of the Product Owner, or worse yet the PO is absent and things stall. Senior Engineers get marginalized, since they do things that can take longer than a sprint or have to do with internals not directly related to business needs. And Engineering is about a lot more than features.

Agile works for some development groups. But it definitely does NOT work for many others.

> The OP makes real points about the failings of Agile.

The OP makes some real points about failings of management that predate Agile, and are common even where the word "Agile" isn't used, and occur (among other places) in places that use "Agile" as an empty buzzword the same as the same type of organizations would use the name of any other approach that was popular if Agile wasn't around.

But none of it really has anything to do with Agile. Very little of it even has to do with Scrum (which, as an actual defined process rather than an approach to selecting/using processes, is at least subject to process-oriented criticism.) And, perhaps most importantly, no solutions/alternatives are offered.

Agile is an engineering process. It is supposed to solve problems. Those problems remain. Ergo, Agile isn't working. We can be apologists all we want for Agile but it remains true, it often fails to improve things at all.

No, Agile is not a process, engineering or otherwise. Agile is a set of principles which can guide the selection of a process and how it is applied, but it is not a process (indeed, it is fairly expressly a response to the fetishization of process.)

And I am unconvinced by your claim that Agile isn't working, that is, I am unconvinced that the problems Agile exists to mitigate are not reduced in organizations that actually apply the Agile principles in hour they approach work.

OTOH, I do think the "Agile" brand had failed because a while lot of applications of canned process in rote ways -- the thing Agile directly opposed -- is being sold as being Agile methods.

Agile isn't an engineering process, its the the idea of applying problem- and resource- (including the staff on the team) specific engineering to the development process.

That's pretty weak - its not a process, its an idea of a process. Whatever. Applying this idea of a process often fails, obstructs senior engineers and does nothing to enable the rank and file as advertised.

All the points it makes about Agile are utter nonsense.

In what way does the engineering team end up subject to the whim of the PO? The PO needs to provide the details of what the business needs (someone needs to do that, after all), and what the priorities are, but he's in no way in charge of the team.

If the team has no PO for a significant amount of time, that does mean that a vital part of the team is missing, but I'd like to see any dev team build something without any idea of the business needs. It's the PO's job to communicate those needs, and the great thing about Scrum is that he does so as part of the team, rather from some distant ivory tower.

I also don't see in what way Senior engineers get marginalized. If anything, everybody gets treated as senior, because everybody gets ownership, responsibility and decision making power about the project (though in reality, it's of course the seniors with the clearest ideas and tend to end up with most of the control over the team).

And there's generally plenty of room to include stories (even very large ones if they can't be cut up, though they usually can) about internals, infrastructure and R&D on the backlog.

It's absolutely true that it's not for everybody, but your criticism and that of the article don't hold much water.

I think that the PO criticism is this: An incompetent or indecisive PO can cripple an agile process. That's true, but (except possibly in the case of a startup), there's always a non-technical layer of management over you somewhere, and if that layer is incompetent or indecisive, your project is equally in trouble. Labeling that layer "PO" and the process "agile" changes exactly nothing about the problem.

I've been in an agile process. It worked, mostly, but it broke down when the PO didn't have the time to do the work. (It was huge for an agile team - 30 people - which put a huge amount of pressure on the PO position).

For the next project, the PO "didn't have time" to give us actual requirements - just "improved reporting". In the most amazing display of agile that I've ever seen, the development team lined up a series of customer conference calls to expose their requirements for reporting, and proceeded to run the entire development process without a PO. And that worked amazingly well...

... right up until three weeks before the scheduled end of the project. (We were probably going to miss by a week, on a six-month project.) Then product management showed up and said "We like what you're doing. But we need these additional things as well. Oh, and we can't change the end date." Predictably, that was the end of agile, as well as the end of any management reality for that project. (You think you can add three months of work without changing the end date? The real world will demonstrate to you that you cannot in fact do that.)

30 people in a single team? Yeah, that's way, way too big. I've been in some teams that were technically too big, but those were still less than half that.

Still, good job making it work despite utterly dysfunctional management there. Agile does require that management understands what you're doing and what their role in that is. But again, that's true in any other situation too.

Well, it worked really well for the release before that. It took a really strong proponent who was willing to hack process as we went along, but it worked - for us, for a while.

I'm thinking we have very different ideas of what a Senior Engineer does. Or how capable a dev team can be when given a project, without micromanagement from a PO about which buttons go where. And whether its reasonable to cut a big project up into hash and try to (appear to) complete something useful to a customer every 2 weeks.

Maybe we do. I don't know what you think a Senior Engineer does. I think a senior should be able to take responsibility and ownership over his work and the work of his team. I don't think they're able to predict business needs without any feedback about those business needs. I certainly think a senior should be able to divide big chunks of work into smaller parts, because even a junior should be able to do that; that's pretty much the core of programming.

I also think a senior should be able to understand how a development process works without resorting to silly strawmen.

There are a multitude of processes for getting feedback; Agile is a contender but by far not the best. Please don't be disingenuous - Agile processes are often strict and brittle - anything that doesn't fit into the Agile round hold can't be done, and the 'sprint' is the worst of the awkwardness.

Strawmen are different from examples; its lazy to call any explanation a strawman.

In what way is Agile strict and brittle? It's entire point is to adapt quickly, on short iterations. To be flexible. In probably the most strict form of Agile, Scrum, constant improvement of the process is an intrinsic part of the process.

I'd like to see a process that handles this better while clearly being not Agile.