Confidence vs. Competence

| 1 Comment

Tech Competence: How good you actually are at whatever it is that you're doing.
Tech Confidence: How good you think you are at whatever it is that you're doing.

This post is inspired by a recent article that made a lot of very good points relating to the grass-roots of Open Source projects. As it happens, the same points also apply to closed source products (such as what I do for a living) and to the Systems Administration community as a whole.

For the tl;dr crowd, a bulleted list of key points:

  • Building a culture of elitism (also known as a robust meritocracy) means:
    • You risk not being able to replace retiring gurus with fresh ones.
    • You risk not being able to grow the community as a whole due to the barriers in place to keep out the non-elite.
    • Efforts to change up the diversity in the community will largely fail due to those same barriers.
  • Communities defined by tech competence are more often actually based around tech confidence.
    • Building barriers based on competence therefore do more harm than good in sustaining that community.
    • To grow such communities, it is far more effective to focus on building new member confidence.

For fully volunteer efforts like an Open Source project, these are very key things to keep in mind. If a project gets too cliquey it risks dying from lack of investment. Such projects live and die by their community.

For professional efforts like what I do for a living (we get paid for this so we kind of have to work together) you can get away with not doing this kind of thing, but it's still a bad idea. For one, new-hires will have to be shoe-horned in somehow and if the going gets rough they'll just jump ship to a less difficult company. As always, office culture matters quite a lot.

For outreach efforts advocating for a specific community it also matters quite a lot. For instance, LOPSA is such an advocate for the Systems Administration community as a whole and has faced significant headwinds in part from Systems Administrators generally being perceived as an elitist bunch of assholes (the dragon in the datacenter; do not poke unnecessarily, or your User Picture will be turned into smoking boots ). Changing an entire profession is a very hard problem.

Now for some specifics.

My Office

We're a closed-source office so these guidelines aren't nearly as critical to survival as it is for the OSS side of things. That said, we work on an obscure enough problem that we can't write job-postings that ask for just what we're looking for and expect to get useful candidates. We'll get resume-shotgunners. We know, we tried.

Must have #{min_years_dev} years in #{dev_env}.
Must have #{min_years_market} years in document processing, including: #{doc_list.join(', ')}

We'd be extremely lucky to get one shining example, who probably wouldn't mesh with the office culture for some reason. And then we'd have to let them go.

Good thing we actually value confidence! If someone shows enough gumption in the right technical direction for us, they can pick up the esoteric things we do that aren't that common. To do that, we can't be the kind of office that values the Secret Knowledge, or are dicks about working with people who don't yet know what they're doing.

Having silos of knowledge is a hard problem to solve (Brandon is The Man when it comes to Chef problems, so he gets all the tricky ones; except for when he isn't there, when his number two has to step in and learn a lot in a short period of time. Hooray confidence), but we do what we can to make sure more than one person can solve problems related to each technology. Or at least knows enough to be confident they'll get to a solution if they have to. The office culture supports this as well through several of the methods used in the article for the Dreamwidth project.


ServerFault is a community of professional Systems Administrators out to build the best Q/A resource of sysadminly knowledge on the Internet. We've done quite well, as our pages rank very high in Google results. However, all is not well.

  • Our guru replacement rate has dropped below the guru retirement rate.
  • Our remaining gurus are feeling besieged by those who don't meet their standards for membership.
  • Our outreach efforts haven't really worked.

These are strong signs of a meritocratic/elitist system, and they're right. ServerFault is unique in graduated StackExchange sites in that we're defined by professional membership (there are a couple of beta sites who are doing the same, though), which means we're actually exclusive. We've spent our entire existence trying to walk the tightrope of minimum technical standards versus open access, and it has been a hard, long, ongoing discussion with the community.

The thing is, we've recognized that confidence is something we look for. Confidence in the sysadmin space shows in evidence of prior research before asking on a Q/A site. That's tricky to codify into our definitional documents, but we've done so. Questions like, "give me a step-by-step install guide for $product in $OS using $framework," don't show evidence of 'confidence', or even 'competence'.

By being a 'professional' space we're cutting ourselves off from our own grassroots. This was done intentionally, since experienced sysadmins get ever so tired explaining over and over again how to do some basic thing. The hope was that by being an attractive place with a good signal to noise ratio, the professionals already out there would start using us as a resource and clearinghouse.

There are three levels of competence at work here:

  1. I know enough to answer this off the top of my head (what the article writer deemed tech competence)
  2. I know enough to phrase search queries effectively to locate an answer (what the article writer deemed tech confidence).
  3. I don't know enough to phrase search queries effectively, so I need hand-holding or an interactive session with an expert.

ServerFault is all about the first two, and very much not about the third. People in both tiers form our cadre of active answerers. It's the questioners that we're having big problems with; these days they're mostly coming from the third set, and that doesn't pass either competence or confidence tests.

Systems Administration in General

Some of ServerFault's problems can be lain at the feet of the long-dominant culture of social sysadmins. Sysadmins have been on the Internet since the very beginning, so there is a long tradition already in place that precedes the non-ASCII revolution (also known as this web thingy). One very common theme in nearly all sysadmin spaces is this sentiment:

We are not the Helpdesk for the Internet.

This strongly ingrained not value has the effect of cutting us off from our own grass-roots. A lack of training programs that produce Sysadmins who pass the competence smell-test means that nearly all of us came at this profession through the grass-roots somehow. Some of us had mentors to guide us through this transition from newb to full blown member. Some of us did it all in read-only mode, never asking questions on the sites we kept running into on Google. Some of us haunted IRC, asked questions, and weathered the shit-storms that created.

And yet, even professionals who pass the confidence tests in spades and have the technical chops to back it up sometimes bounce right off of these sysadmin spaces and continue not being social about it. A culture of exclusion causes this.

  • A 15 year Windows admin is looking to do some Linuxy administration tasks to try and freshen up their resume. But doesn't engage on the project-support mailing list for fear of being called out as incompetent and lacking key knowledge that everyone knows.
  • A grizzled veteran of at-scale IIS web-serving looks to get into Apache, but doesn't actually talk on the channels because of the outright scorn they see directed at IIS.
  • A 10 year Linux veteran who has been doing Python and Shell scripting their entire career is thrust into a shop that only does Perl. Since everyone knows Perl, obviously, they don't ask for assistance in getting up to speed.

These are people that would be right at home in their home tech, and probably have some Tier 1 knowledge areas in them. Yet, they're Tier 3, or maybe just getting to 2, in the new one. They can tell they don't have the Secret Knowledge yet, even if they only think they don't, so go to do more self-learning rather than engaging.

Fixing This

The problem in the sysadmin space is cultural, and deeply ingrained at that. Changing that kind of culture is done at a glacial pace, to the great frustration of those who desire to change it. LOPSA has been trying to recruit more Windows admins into their fold, but it's very hard going; in part because our own mailing-lists are full of people who say thing like "Garg, Windows makes no SENSE," or outright mock how things are done in that environment.

There has been a lot of talk in recent years about building degree programs for what it is that we do. They're out there, and the quality of them is a topic of much debate. The people running the programs are involved in the debate and are actively working on improving what they do. Building an on-ramp into the Tier 2 level where community acceptance is a lot easier to achieve is a very nice start, but we still need to find a way to welcome self-learners if we're going to improve the reputation of the profession as a whole.

1 Comment

Excellent write-up, as usual. Just wanted to drop a few thoughts on you:

When you first mentioned tech confidence and competence I assumed you would mention the Dunning–Kruger effect (whereby newbies know they know nothing, intermediates think they know everything, and experts know about all the things they don't know). It strikes me that the System Administration field falls victim to the D-K effect most of the time (I can admit that I feel into that trap early in my career). At the same time, that "problem" is solved handily by mentoring and exposure to the full spectrum of administration. I mention this primarily as mentoring solves more issues than what you've discussed and it's importance can't be understated.

As for the lack of mentors in general, I think to how I was saved from the D-K effect. In my case I started reading, veraciously, and still do to this day. When I encounter someone who is just getting into the field, I wonder how I could possibly make any notable impact on their knowledge. It would be as encountering a mountain of sand and a shovel - spending a whole day toiling yields no visible difference. On a website like Server Fault, it's the same, but there's a new mountain of sand every 15 minutes.

I would like to see the culture of system administrators change too. I'm not sure how to enact such change, as there are more factors which combine to keep the status-quo.