Finding inspiration in the world around us

Episode 18: 07/07/2021

?

?

When I was about twelve, our school assemblies were led by the headmaster, Mr Coatham. It was a mildly religious school, so some of Mr Coatham’s assembly talks were stories from the Bible. But many were just random anecdotes from life.

?

One such talk, which will stay with me forever, concerned a holiday Mr Coatham had recently had with his family. While in northern France, they visited one of the cemeteries from the First World War.

?

If you were a man of conscriptable age during 1914-1918, you were highly likely to be loaded onto a train and taken to the front lines, which basically ran the length of eastern France, from the sea to the Alps. There, you would live in a muddy hole, or perhaps a cave, for a few months, before somebody blew a whistle and you and your newly found friends ran across a patch of mud directly towards a machine gun.

For the men of many countries, particularly France, that’s where the story stops. Those events are forever marked by the war cemeteries, which are both numerous and vast.

?

You stand in the middle of one, and the gravestones seem to stretch to the horizon. Mr Coatham was roughly in his 50s, so his daughter must have been fairly grown up when they visited the cemetery. As his story went, she walked out into the peaceful field of crosses, looked around, and wept.

?

One such cemetery can be found in Arras, which we visited on the train a few weeks ago. The trip was much less eventful than it was supposed to be, thanks to a conspiracy between Google Maps and the local French businesses, which makes many of the attractions in Arras difficult to find, and shut when you get there. The cemetery is never shut, and commemorates the British forces who died but could not be identified and buried individually.

We also managed to penetrate a museum about the vast tunnels beneath Arras. The tunnels are actually a disused chalk mine, and the Allies hid tens of thousands of soldiers in them for a surprise attack towards the end of the war.

?

The visit to Arras provided some useful perspective.

We are now in summer 2021. I have been working on El Toco for five years. Every piece of it, from the crawler to the website, has turned out to be much more time consuming than expected. I don't know of any other small startups who have spent this long on their pre-launch phase. Every minute, of every day, is haunted by the dual nightmares of working on something for so long with no payoff, and of the possibility that the payoff never arrives.

?

Junior colleagues, who I left behind in my previous job, are now in more senior roles than the one I used to have. They drive nice cars. I couldn’t possibly afford a car, and put a certain amount of effort nowadays into ensuring we get the cheapest combinations of biscuits in the weekly food shop. Some time during the pandemic, I seem to have crossed over a careers Rubicon. Finance headhunters stopped cold calling on LinkedIn. There is no way back.

?

It is, simply, scary. But fear is not going to stop us. Bravery is about being scared and getting the job done anyway. Above and below ground, the sombre landmarks of Arras are a testament to this. What we face is nothing, compared to what those brave men went through, who are now lying silently under the green grass.

?

?

Picking a business model

?

We haven’t discussed how El Toco makes money in this blog until now. That's because it doesn't. Hahaha. I mean how it will make money, when it goes live. For a search engine, there are basically two options. I initially chose the wrong one.

Money is creeping onto the radar for the second time in this project. At the end of 2020, with the crawler done, it looked like we would launch in short order. But when we go live, all our dormant servers are switched on, and our monthly overheads shoot up.

?

The launch first, monetise later approach, while straight out of the Silicon Valley playbook, is no longer realistic. We will run out of money before we ever get enough users. So, reluctanty, I have brought forward the revenue-making part of our website, which means doing it now.

In an instant, our imminent launch vanished. In its place is another long development slog. We started this new slog at the beginning of 2021, and will talk about today.

??

But first, how does El Toco actually make money? One way is to charge users. This was my original plan. We’d have a subscription service, costing something pathetic like ￡1 a month for all your devices. I spent a few days back in 2016 making a browser plugin for Chrome and Firefox, so that you didn’t have to log in with a username and password.

?

It took precisely one after-dinner conversation to junk this business model. One of my friends made the simple observation that user authentication is a huge faff. Each time some silly website or app asks you to log in, forgets that you have logged in, or just randomly stops believing you are you, it’s extremely annoying. Bang goes another ten minutes of life on a tech-related task which didn’t exist for most of us only a few years ago. Even if my browser plugin worked, it was bound to annoy somebody.

?

Since Google and the other big boys can be freely used, a competing platform is at an immediate disadvantage if it resides behind a login. At least one other search startup is already trying this. It is called Neeva, as in "neeever going to work, Mr Bond". Which we will see, in the fullness of time.

?

Neeva

?

Some time during the pandemic, the story of Neeva emerged. Two senior executives had left Google and decided to set up a rival search platform. The news sent chills down my spine, as it does any time I read about new search engines.

Fortunately, Neeva didn’t look like a threat. The unconvincing copy on their home page couldn’t provide any concrete reasons to use them. Their service seemed unavailable in France. But the fattest nail in the coffin was that they were charging users.

?

Neeva will keep going until 2023, whereupon they will fold. They won’t mention any redundancies, so on the day this news breaks I will listen guilt-free to the song “Another one bites the dust” by Queen. On repeat. You have to savour the small victories. Ostensibly, the reason they will fold is that their team of twenty five want to work on artificial intelligence. Whatever that means.

?

My suspicion, however, is that they had both an undifferentiated product and hampered it with a business model which doesn’t work. Possibly they had to choose this business model, in order to evade some non-compete agreement they’d signed with Google. Nobody will ever know except them, and they aint talking.

?

?

The part where we set out to create the advertisers’ section and oh look, it’s a drama

?

All that happens in the future. Back in 2021, charging users still feels like a bad idea. That leaves us with the other option: selling advertising.

Searching on El Toco will be free, supported by sponsored search results. This is our business model.

?

The search section of our site was finished just before the lockdown Christmas of 2020. It looked nice:

?

As usual with this project, there was a downer. This blog wouldn't exist if everything went smoothly, it'd be way too boring to read about.

This particular downer was that it took the website company five exasperating months to create the three pages which make up the search section. Five months of honeyed words in email chasers. When what I really wanted to do was storm their office and force feed them Apple smart watches until they got the idea that time is very important to clients who are slowly going bankrupt.

You may feel that last joke didn't really work. That's actually the aim. Because the three pages that they delivered also didn't work. Don’t think that they implemented the search functionality. Oh no. You couldn't type something into the box and actually get search results. That part is far too complicated to contract out, I had to do it.

?

Now, all of a sudden at the start of 2021, we need to create the other half of our website, which lets advertisers create and manage adverts. It’s considerably more than three pages. We obviously don’t want to reward failure by giving this work to the website company.

?

But we will anyway. The website story, which has played out over the first six months of 2021, is as follows. From my admittedly self-centered viewpoint, this is the second time our launch timeline has been significantly damaged by this website company. You can read the events that follow and judge for yourself.

?

It starts with the spaghetti code.

?

One day in January, I sit down with a large pot of coffee to begin creating the advertisers’ section, using the cheery colour scheme of the search pages.

?

It is nice coffee. This is a good thing, because my only achievement that day is drinking it. Under the hood of the website is a wriggling mass of spaghetti code, which makes very little sense to anybody except its original author. You know, the developer at the website company, who we don’t want to use any more.

?

Crikey. We are already in a bind, after just one day on the job. Unpicking the spaghetti code, so that it can be generalised across the whole website, will take weeks. During those weeks, I will be prevented from doing anything else, because I’m an inexperienced web developer so these things take me a long time. There isn’t a proper design, because I’m also an inexperienced web designer. Not only will this use all my time, the result that pops out at the end won't have a cat in hell’s chance of looking professional. It's going to suck.

?

Very reluctantly, I go back to the website company. Could they add the remaining pages to our site please. Unfortunately, they can’t. They’ve sadly furloughed their team.

It could be argued that, had they done our work quickly in 2020, we would have awarded them more work, which could have helped this situation. But our contracts don't cost much, so it may not have been enough. The long shadow of the 2020 pandemic is bad news for them and also bad news for us. Now we need a new website team.

?

I find another freelancer. They start work, patiently applying the search section styles to my crude wireframes, which have progressed from being in Microsoft Word to being in Microsoft Paint. Anybody reading this who has actually worked in website development will have their head in their hands after reading that sentence. One does not design websites in Paint. I will learn to use professional design tools, but that’s still a few years from now.

?

The freelancer has been working for two weeks when the original website company gets back in touch. They’d like to unfurlough two employees so that they can work part time on our project. Tough luck, I can’t mess the new guy around like that.

?

Over the months that follow, the advertisers section gradually rolls off the freelancer’s production line. This is how it looks:

?

As before, none of it works, the freelancer is just doing the pretty visual stuff. It’s an improvement on nothing, which is what we had before. But you can perhaps see that the designs are starting to look a little pedestrian. Nothing screams “this is a cutting edge product you’ll love to use”.

?

While I work on the complicated back end, which we’ll come to in a second, I conclude that we need to polish the design. We also need to add some landing pages, because right now there’s nothing after the main home page to explain why you’d want to advertise with us.

?

The freelancer says that they’d love to make these changes, but they can’t. They’re going on holiday for basically the whole summer. There’s nothing morally wrong with that and indeed I envy their work-life balance. By this point, our grad has resigned, so cash burn has dropped almost to zero. But when we are literally years behind schedule, burning yet more calendar time, waiting for the freelancer’s jolly summer holiday, is unattractive.

Another day, another tricky call to make. Do we find a third website contractor, mid-project, to pick up the work?

?

I decide that finding somebody else is the wrong route. Too many cooks spoil the broth, and the code is already degenerating as more people work on it, making it steadily harder to maintain.

?

Back we go to the original website company, to ask they still have capacity to work for us. Happily, they can and we agree a reasonable rate.

?

Then comes a little catch. Neither the freelancer’s code, nor their designs, are worth keeping. Just weeks after is was delivered, everything the freelancer spent four months doing for us is going straight in the bin.

Then comes another little catch. After we pay them up front, it emerges that the original developer, who worked on our project, isn't employed there anymore. A third cook is going to be working on the broth anyway.

?

That is how things have played out up to summer 2021. In future years, I will come to consider this the second strike of the website company on our launch timeline. Obviously, none of this was intentional on their part. Obviously, decisions taken by me were what allowed the damage to happen. But, had they been available at the start of the year, we would have gone with them directly. Instead, we went with the freelancer, wasting a third of a year and several thousand pounds, to build what Italians call a cathedral in the desert. It was not a great day in the El Toco office when all that sunk in.

?

You might expect this has triggered a period of intense introspection, asking if this reveals some failing in me as a manager.

?

Perhaps it should have, but there isn't time for any of that. While the freelancer did the pretty parts, I’ve been working on the back end of the website. It has turned into one of the most fiddly parts of the whole project.

?

?

Back end frameworks

?

?

During the early 2000s, jeans were ultra low rise. A necessary result of this was them slipping any time you sat down. This particularly afflicted girls, because fashion dictated that they also wore tops which ended somewhere around the belly button. There was really nowhere to hide as the two garments stretched inexorably in opposite directions. Irrespective of whether you were a plumber or a supermodel, it was always a bit of a naff look. As the decade drew to a close, everybody switched to skinny jeans, waistbands crept up, and it was generally agreed that this was a more aesthetic and comfortable style.

?

Except that of course it wasn’t. If they are not well elasticated, skinny jeans are very restrictive, which was particularly apparent any time you wanted to put them on or off. To overcome this, women developed a sort of twisting hip thrust which would not have looked out of place in a Caribbean dancehall music video. If you happened to have a proportionately narrow waist, a new problem developed, which is that the jeans fitted just fine lower down, but had a draughty gaping waistband at the top, generally getting in the way and suggesting that you were wearing the wrong size.

?

Jeans have since switched back to a looser fit, but without the low waistline of the early 2000s. The underlying logic is apparently practicality and comfort. Visually, these retro styles are unflattering for all body types, but that is precisely the point.

?

But what is the actual point here, analysing jeans in an article about websites?

In 2001, if you had bought ten pairs of high waisted jeans, which were slim but not tightly fitted, you could have sailed through the last twenty years and never had to throw a pair of jeans away before they wore out. That is the actual point, because this is just like programming.

?

?

Programming trends come and go. If you build your company’s systems around something which turns out to be a fad, you’re in trouble. Because when it becomes obsolete, you’re stuck with a wardrobe of expensive skinny jeans which nobody wants to work. In the IT world, this means rebuilding everything from scratch.

?

My favourite example is, as usual, a Microsoft product. This one is called Silverlight.

It was some sort of web programming tool, but that doesn’t matter any more. All you need to know is that now, if you search for Silverlight, you will find a page from Microsoft’s official website, explaining in the present tense how it’s the next big thing in web design. Next to that, is a page from Wikipedia, explaining that Silverlight will be discontinued in 2021, because nobody uses it. Anybody who delivered their… whatever using Silverlight is presumably in a bit of a pickle.

?

Coming back to El Toco, while the freelancer works on the user-facing, visual bits of the website, I been doing all the back end functionality. So that when you click "log in", it actually logs you in.

Unlike the search section, the advertisers section of El Toco's website is very similar to other, normal websites. There is a login area, an order flow, an account settings page, etc. Coding this all by hand is reinventing the wheel. There had to be a better way.

That better way is to use what they call a back end framework. There are several competing ones, so this first involved wading neck deep into the freezing waters of programming trends, figuring out what all the made up words mean, and deciding which framework we're going to build our website around.

?

I picked the website framework by following along with some “let’s code together” videos on YouTube. Back in the good old days, I always used books for this. Worryingly, there are very few books on website frameworks. It really underlines how ephemeral they are that nobody even bothers to publish books about them. Here one day, gone the next.

Watching the YouTube videos felt like doing a correspondence course in web development. Only with ridiculously high stakes, because our whole company depends on the website being done properly this time. Some of the frameworks involved much more faff than others, demanding that you learn and understand a whole lexicon of internal jargon. I'm sure they're great but they are clearly designed for large teams, who manage complex websites like ASOS. El Toco needs simplicity. I chose a framework which seemed simple.

?

That took just a few weeks. Then, I had to actually create the website. That step has consumed the entire first half of 2021.

?

Actually using the framework is like building with Lego. The freeform Lego, without the instructions. You get all the components, but how they’re strung together is entirely up to you.

For example, the framework provides a way of checking if a user’s password is incorrect. But you, as the developer, has to decide what happens then. How many tries do they get? Do you lock their account? Send a notification email? Which system sends that notification email?

?

Before long, you will have spent one calendar month, just handling incorrect passwords. It's the same deal with the part of the website allowing them to place orders. Only worse: when you think of all the things that can go wrong just while taking credit card payments, there's a lot of scenarios to handle.

That is how I've spent the first half of 2021. Six months, building a robust log in and order flow. Since El Toco has been running for five years, this means 10% of the whole history of the company has been spent on the back end of our website. I wouldn't say it's gone badly, indeed this is the system we will use for our eventual launch. But it is a fat chunk of time. Dedicated back end developers must surely do it faster but, at this late stage, we couldn't possibly afford one.

?

?

Spooky noises from a distance

?

Air raid sirens go off on the first Wednesday of every month here. It is unsettling.

?

The first time it happened, I had only ever heard air raid sirens in museums. It’s a few decades too early for the terminators, so my initial thought was that I would spend the last seconds of my life wondering which inept, power-mad leader had pressed the nuclear button. There were so many to choose from.

?

My next thought was to look it up online, where I discovered the official explanation, which is to ensure the sirens work, in case they are ever needed again. Like the vast cemeteries in this area of Europe, it is another reminder of how trivial our daily woes are, in the scheme of things.

?

That perspective will be something to hold on to during the second half of 2021, when the website company resumes work. But that’s a story for another day.