Every company still has a webmaster

The man whose job was to say no

I started near technology by accident. I did an internship as a media designer, which in those years meant you sat close to the people who made the website without ever being allowed to touch it.

I had ideas. I took them to the marketing department. And then they had to be built by somebody else, because the person who ran a website was called the webmaster, and he was the only one who could.

Every single time I asked him for something, he took my request and turned it into a kind of mysterious magical protocol, and then said no, using words I did not understand at the time.

I want to be precise about what that felt like, because it is the part companies never see. It was not that the answer was no. It was that I could not tell whether the answer was no because the thing was impossible, or no because he did not feel like it. I had no way to check. Neither did anyone above him.

He was not the villain, and that is the whole point

It took me years to understand this properly and I want to say it plainly, because the easy version of this story is wrong and the easy version is the one most people tell.

He was not a difficult man protecting his territory out of spite. He was one person holding the only key to something the whole company depended on, with more requests arriving than any person could serve, and no way to explain a technical refusal to people who had no vocabulary for it.

Of course he said no. No was the only answer that scaled.

The bottleneck was not his personality. It was the design. Somebody had decided, probably without deciding anything at all, that one person would hold all the access, all the knowledge and all the judgement about what was possible. Everything that followed was arithmetic.

He was not a difficult man. He was a bottleneck with a desk, and the bottleneck was the design, not the person.

What I did about it, which was not clever

I went and learned it myself.

Not to replace him. I could not have replaced him and I did not want to. I learned it so that I could tell the difference between the two kinds of no.

And then something changed that I did not plan and did not expect. I stopped arguing with him about whether a thing could be done. I started arriving with the answer already written. Before I presented an idea anywhere, I had gone and found out whether it was possible.

Which meant we stopped fighting. I would pull a chair up to a second computer beside his and show him why it was possible, and he would tell me the thing I had missed, and both of those were useful.

That is the moment I fell in love with computers. Not at my first line of HTML. At that second computer, working with somebody instead of asking them.

What it cost the company, which nobody counted

Here is the part that should interest anybody running anything.

I was one person in a marketing department and I went and taught myself a technical discipline in my own time, in the nineteen nineties, because that was the only route I could find to a yes.

Most people do not do that. Most people ask twice, get a no they cannot evaluate, and stop having the idea. They do not escalate it, because escalating means saying you do not understand the explanation you were given. They do not go around him, because there is no around. The idea simply does not come up again, and nothing anywhere records that it existed.

That company never knew what it did not get. There is no line in any report for the ideas that died in the asking. The cost of a bottleneck is invisible by construction, which is exactly why it survives.

The job died in 2003, and the role did not

WordPress arrived on 27 May 2003, and within a few years a person with no technical skill could publish without asking anybody. No file transfer, no waiting, no protocol.

The job of webmaster genuinely went away, and I have never pretended that gave me nothing.

But the role came straight back, and it came back because it was solving something real. Somebody does have to care whether the thing is secure. Somebody does have to say no to the request that would break everything. Gatekeepers exist because gates are useful.

So the question was never whether to have one. It is whether what you have is a gate or a wall. A gate opens for the right answer. A wall is the same height no matter what you bring it.

Where yours is sitting right now

In the companies I look at, the webmaster is usually in one of four chairs.

On the data. One team can answer questions about the numbers and everybody else submits requests. Nobody outside that team can tell a hard question from an unwelcome one.

On deployment. Anyone can write the change. One person decides when it reaches a customer, and the queue is measured in weeks.

On the AI tools. This is the new one and it is moving fast. Access to a capable model, or to the systems it is allowed to touch, sitting with whoever got there first, handing out answers rather than access.

On approval. No technical gate at all, just a person whose yes is required and whose calendar is full. This one is the hardest to see because it looks like process.

In every case the test is the same. Ask the people one layer out whether they can tell the difference between that is not possible and I would rather not. If they cannot, you have a wall, whatever the org chart calls it.

What actually fixes it

Not the tool. I want to be clear about that, because I sell this for a living and it would be convenient for me to tell you otherwise.

The tool arrived five years after I needed it. What fixed my situation was that I became able to check the answer, and the company could have given me that in a week by letting me look. They did not have to teach me anything. They had to stop treating the knowledge as the gatekeeper's property.

So three things, and none of them is a purchase.

Make the refusal legible. A no should come with a reason a non-technical person can evaluate. Not a protocol. A sentence.

Give read access much wider than write access. Most of what a bottleneck protects does not need protecting from being seen. Being able to look is what lets somebody arrive with the answer already written, and that single change turns a queue into a conversation.

Count the asks that stopped. Not the tickets, the asks. Ask three people what they wanted last quarter and did not pursue. You will hear things that nobody ever logged.

And if you are the gatekeeper, which plenty of good people are without choosing it, the move is the same one he eventually made with me. Let somebody pull a chair up. It is the only version of this where you stop being the person who says no for a living.

Why this is the work we do

Almost every engagement that starts as a technology question turns out to be this. Not which model, not which platform. Who is allowed to find out whether a thing is possible.

The organisations that got the most out of every tool of the last thirty years were the ones that widened that circle. The ones that bought the tool and left the gate exactly where it was concluded the tool did not work.

The tool worked. Nobody was allowed to use it.