返回资讯中心
外部精选
软件工程
#开发者实时流

The coastline paradox will kill your startup

Welcome to the 8th issue of the o0 Letter - we are here to help you get from Seed to Series B so fast that everyone will wonder how you did it (tell them we helped, please).

Hacker News 最新tosh2026/10/20 · 26 天前16 分钟阅读

以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。

o0 Letter

Oct 18, 2026

5

2

2

Share

Welcome to the 8th issue of the o0 Letter - we are here to help you get from Seed to Series B so fast that everyone will wonder how you did it (tell them we helped, please).

Most problems are coastline paradoxes. From the zoomed-out position, they can seem like easy things to do, but the more we zoom in, the more complex they get.

What is the coastline paradox? It comes from the fact that, if we were to take a to-scale map of Great Britain and put it on our desk, measure it, and then use the scale to calculate the length of that coastline, we would come up with a length of the coastline; but if we made that map bigger, the calculated length would get bigger, even though it is to scale.

If we made the map the scale of the room we are in, there would be so much more detail and so many more crevices that it would significantly add to the length of that coastline. If we went all the way to making it as big as the real coastline, we would find that it is significantly longer. Lewis Fry Richardson first studied this around 1950. Measure Britain with a 100km ruler and the coast comes to about 2,800km, but halve the ruler and it grows to about 3,400km.

A startup, and its plans, behave in the same way. From a distance, launching in a new market is like measuring where we want to be with a 100km ruler. The closer we get to understanding what it takes to get there, the longer that line gets and the more work is created. Using the example above, we plan with a long ruler but build with a short one.

I have helped many startups where this has come up. They add what seems like a subtle thing, and then open up a can of worms:

Selling subscriptions, then adding ads as another revenue stream.

When we are a very capable, hard-working and intelligent team, we think, “Oh yeah, we can do that,” and we can, but the real challenge is that the more we zoom into that problem, the more we understand our market, our customers, and the real challenges ahead, and the more we understand the potential model, the more complex it becomes, the more unknown there is, and the more we lose focus.

One startup I later helped was selling SaaS CRM software to SMEs and startups. They flippantly wanted to also sell to enterprise, as a sales goal, because they wanted larger revenue and larger ticket sizes.

Their logic was simple - enterprise software in their niche was not as good as theirs, they wanted more revenue, and if they also sold to enterprises alongside SMEs they would get bigger, faster results.

So, they set a goal to sell to enterprise, gave it to the sales team and started executing, telling salespeople to reach out and sell.

But they didn’t get those enterprise sales. When the salespeople reached out, it was mostly just apathy.

After months, they realized it was way more complex and narrowed the focus back to SMEs and smaller organizations. Ultimately, a lack of progress.

As much as the central need of enterprises was the same, they had not taken into account the immense complexity of the features needed for enterprise software, or the level of work, time and detail required for a process of selling to them.

They ran into enterprise-grade security, on-premise needs, a long list of feature requests (e.g. SSO), long sales cycles, a significantly more complex sales process, different landing pages, different positioning, different ads, and different places to find them (events, etc.). In essence, a whole world of unknown product, features, sales, marketing and positioning was needed.

Say we took the real length of the UK coastline, calculated by literally measuring it in person, and worked out how many days it would take to walk at eight hours a day and four miles an hour. If we were to set off on our trip expecting it to take that many days, it would take significantly longer. Sometimes things will go wrong. Sometimes we will leave something back at the campsite and have to go back and retrace ourselves. Sometimes we will be injured or feeling tired and work slower. Sometimes we will have to make a detour because we did not have enough supplies and have to go off track and come back. Sometimes there would have been erosion of the coastline that meant our way is blocked.

“The first 90 percent of the code accounts for the first 90 percent of the development time. The remaining 10 percent of the code accounts for the other 90 percent of the development time.” - Tom Cargill

To solve this, our default logic should be to assume that by adding a new sector, feature set, geography, market, niche, etc., we will run into the coastline paradox. Before executing, then, if we must, our first step is to “Explore” the depths of the challenge.

There have also been times when I have been in a room setting goals with startups, and there are three or four theoretical pre-product-market-fit theses, all unproven at the same time. They are not trying to whittle down between them, even though each one has the possibility to be a $B Startup. Instead, executing all of them without realizing that they are coastline paradoxes.

99% of the time, things are much more complicated than they look. So it really does mean that we have to be very selective and ruthless in picking the things we do. More on that here.

Have you added goals, projects or targets that could be coastline paradoxes, and which one do you need to prioritize?

Benoit Mandelbrot’s 1967 paper How long is the coast of Britain? explained Richardson’s finding: the shorter the ruler, the longer the coast.

Michael Wolfe’s famous answer on why software estimates run two to three times over: a 10-day coastal hike from San Francisco to Los Angeles that took 60.

Joe Morrissey, who ran revenue at Segment as it moved upmarket, says the biggest mistake is mistaking initial enterprise interest for product-market fit. Segment’s enterprise sales cycles could last as long as 9 to 12 months.

Jason Lemkin warns founders not to use moving upmarket to hide slowing customer growth, and to keep fighting for their smallest customers.

Bent Flyvbjerg’s database of 16,000+ projects finds only 8.5% come in on time and on budget, and just 0.5% also deliver the benefits promised.

Andrew Crump and Nick Black Co-authors, the o0 Letter

PS. If this resonated, forward it to someone who needs to hear it.

正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。

打开原文
去社区讨论 →