My first job out of Uni was a company that ran financial services, logistic fulfilment, and retail. They had at least a dozen large in-house retail brands, several large warehouses with automation that Amazon was probably behind at the time, and they also ran serviced some of the most well known brands in the country - from their website to customer support to logistics. They had delivery networks, networks of direct shipping, vast call centres with thousands of staff. We integrated with the banking system, with all sorts of external companies.
Our total headcount was under 10 000. Most of that was logistics and call centres (taking retail orders). Our IT dept was literally about 200 ish. All our office staff fit within a few floors of a medium sized building. IT was less than a single floor.
I’d say our software, whilst serving fewer users, was completing tasks orders of magnitude more complex than Twitter. We didn’t have much down time, in fact almost zero beyond physical damage to our networks and the once every few years disaster. Our critical support team was a handful of greybeards who spent much of the time playing solitaire. Code reviews were tight and access to production strictly controlled.
What we did have was a very lean culture, carefully managed over a long time. The company was over 100 years old.
I do not understand how, with far more sophisticated tech, companies like Twitter need so many staff. We could have run Twitter with a thousand folks, including sales. From an engineering standpoint a few hundred would do. And most of us weren’t great engineers. There were a few very experienced freelancers, though. We all worked 0900-1730, but we worked 7 full hours every single day, every minute of those 7 hours. No beanbags. Most of that time was sat quietly coding in a very cleverly designed open plan office (no cubicles) that was like a library.
Looking back, I think the key insight was that we were very cost driven and absolutely kept everything as simple as possible. We had all the usual project overrun problems, usually because things turned out to be more complex than previously imagined. But it baffles me when I hear engineering head count at places like Twitter. If you had given us 1000 engineers, we wouldn’t have known what to do with them.
It is amazing how quickly complexity comes about to serve complexity. Like at $CURRENT_JOB we split everything into a few dozen microservices with their own DBs. Turns out that we need all the data in one place to make some decisions. Shocking I know.
Cue a team of engineers working for a year+ to build a complicated data pipeline to bring all that data back together into a graph DB (not a clue why a graph DB) and build a DSL to write the code to make these decisions.
That's where the engineers go. It's complexity that is introduced setting off a chain reaction of increased complexity to account for the complexity introduced. Then you need complexity to account for that complexity and so on.
Not that it's all self inflicted wounds. Running a site that counts 1/10 of humanity as active users is a hard thing to do
Our total headcount was under 10 000. Most of that was logistics and call centres (taking retail orders). Our IT dept was literally about 200 ish. All our office staff fit within a few floors of a medium sized building. IT was less than a single floor.
I’d say our software, whilst serving fewer users, was completing tasks orders of magnitude more complex than Twitter. We didn’t have much down time, in fact almost zero beyond physical damage to our networks and the once every few years disaster. Our critical support team was a handful of greybeards who spent much of the time playing solitaire. Code reviews were tight and access to production strictly controlled.
What we did have was a very lean culture, carefully managed over a long time. The company was over 100 years old.
I do not understand how, with far more sophisticated tech, companies like Twitter need so many staff. We could have run Twitter with a thousand folks, including sales. From an engineering standpoint a few hundred would do. And most of us weren’t great engineers. There were a few very experienced freelancers, though. We all worked 0900-1730, but we worked 7 full hours every single day, every minute of those 7 hours. No beanbags. Most of that time was sat quietly coding in a very cleverly designed open plan office (no cubicles) that was like a library.
Looking back, I think the key insight was that we were very cost driven and absolutely kept everything as simple as possible. We had all the usual project overrun problems, usually because things turned out to be more complex than previously imagined. But it baffles me when I hear engineering head count at places like Twitter. If you had given us 1000 engineers, we wouldn’t have known what to do with them.