This is a sincere question. I would love to know what the 4000+ staff do there, running a forum web site that uses community moderators.
Correction: 2000 staff. Question stands.
Correction: 2000 staff. Question stands.
Usually when an engineer asks this question, they are starting with the assumption that most everyone who works there is an engineer. That assumption is usually wrong.
I have no inside information at all, but my guess is that engineers make up less than 1/3 of the employees. I'm sure the sales team is large, the product team is large, and the community management team is large. They probably also have a bunch of legal, and we know they have a lobbying group as well. And of course if you're taking people's money, you need a bunch of customer service to handle that.
When I worked there 14 years ago, we had at peak six engineers, and three non-engineers (one sales, one product, one community). But also we had the parent company providing legal (and there was a lot of legal requests even back then!), HR, facilities, finance, and basically all the other support tasks. And also we the engineers did a lot of it. We managed customer service for ads, we managed making t-shirts and plushies, we managed customer service for reddit gold payments, and a bunch of other "non-engineering" tasks.
The site is big and is transactional, which makes it a harder problem to solve. With the near-real-time display of votes and listing changes and comments, it's a lot more like eBay than Slashdot (which is what I would say at the time).
Disney isn't going to just pull up a self service portal to buy ads. :)
I don’t know why you’d jump to the conclusion that the sales people at an ad-supported website would sell ads.
Back then, when there were a dozen of you it was easier to reach an admin when needed than now, heh.
I hear ya. But in their defense, the regulatory environment is way worse now, and back then as an admin we all pretty much knew about every ongoing issue. A current admin can't just answer you back without possibly running afoul of either a regulatory issue or an ongoing project in another group.
who are they selling to? And why so many people?
EDIT: sorry, I read another response, they are selling to ad companies. Does that really demand hunderds, over 1000 people, though? At some point I feel they are costing more than what the next increment can bring in.
I understand your point but even if there was only 100 engineers (which would not be enough) I can't imagine what the other 1900 administrative people are doing. It's an internet forum, and unlike Twitter and other social media (where I assume they simply have hundreds of mods) they successfully outsourced moderation to volunteers.
Is it all just legal suits? The customers are the users and admins are horrible at responding to user requests (be it users or mods), so it doesn't feel lik customer service is using those resources.
This vastly underestimates how much moderating reddit is already doing. When I worked at reddit, we spent at least 50% of all of our resources on fighting spam, harassment, and illegal content. I'm sure reddit is still doing this. What the moderators see is already after the core platform has removed the most obvious offenders. Most of what the volunteer moderators do is remove content that they feel is not in line with their community.
> The customers are the users
The customers are the people who pay, who are mostly advertisers. Customers and users are two different groups with different ideas of what makes reddit good, and sometimes those don't align. It's a fine line to walk between serving the customers and users to try and keep both happy.
I likely am, so thanks for giving me an idea. Admins never really talk about how much they do behind the scenes (and aren't that responsive to the users in general) so its hard to tell.
But as a counterargument: While I'm sure they need more admins now than back in 2010, they also (hopefully) will have automated their spam/illegal filters so it doesn't require 1000x the admin moderators to control 1000x the content. the filter catches maybe 95% of the truly illegal content and then admins can check at a less urgent pace.
>The customers are the people who pay, who are mostly advertisers.
I guess it's more accurate to say that advertisers are the customers, and users are the product. But it's an odd metaphor here since sales would need to appeal to both. Getting ads to show and also getting ads for their own site, to get more products to appeal to more customers.
I do agree that its an odd balancing board, especially with Reddit that is more hostile than usual to inorganic content.
This might be true.
> Most of what the volunteer moderators do is remove content that they feel is not in line with their community.
This is certainly not true for my sub. There's a lot of obvious spam that has to be manually removed. For example, if Reddit enforced the selfpromotion ratio rule programmatically, there'd be an order of magnitude less spam to deal with.
---
Unless Reddit cranks up the removal rate, moderators are still necessary or the /new feed will drown in spam. If /new drowns in spam, what reaches the top will become a lot more chaotic.
But RES has stopped feature development in 2021. If it's not there, it's not coming anytime soon.
> It's unfair and unwise to undercut reddit inc.'s development work. Get gold!
This is about Gold but it might be a more general "we do not step on Reddit's toes" kind of rule too.
I bring this up only because I find when I do this blaming of <other non-build people>, I often miss the problem that is actually in front of my face and can actually impact.
From my experience: as a game dev we ALWAYS know when a game sucks. Maybe some director or other director level person strongarmd it, so it's not like engineers can't be blind. But as a general consensus, I've met very few "bad devs" from a technical POV (a vector consumers often like to attack when they see bad gameplay). We work with a bad hand and we almost never get the extra time needed when we try to show publishers. Fortunately, it's a lot easier now to argue for delays than over a decade ago.
I guess this was Elon's thoughts when he laid off most of twitter.
Edit: Found a good answer for Uber on quora: https://qr.ae/py4zBU
are you saying that the Uber-Driver's-App/Backend will communicate with the driver via these different mapping apps? or that so long as they use the Uber-Driver's-App to stay in contact with Uber, that Uber accommodates them also using other mapping software?
Conversely, I would be more than highly surprised if it was. It has all the trappings of an urban myth.
You then need to have a customer support service. That is probably 5,000 employees right there.
Just because a company has 36,000 employees it doesn't mean they have 30,000 people working on the tech. There are lots of jobs that need to be done just to operate day to day. That most people on here wouldn't even think about, like cleaning staff for example. How many people on here even thought about cleaning staff?
I think it's very likely Uber employs cleaners themselves due to costs attached to outsourcing and that they already have the staff to manage that sort of worker.
But 36k still seems absurd.
They now also sell Eurostar tickets (although I don't see the advantage over Eurostars website)
They're sponsored by Uber, that's it - they bought the naming rights. The actual boats are owned and run by the Thames Clipper company.
Their stated goal to supplant as much of the existing TLC infrastructure for the planet as they can get their hands on. You can bet that these organizations (even restricted to those markets they are successful at "disrupting") employ a lot more than 36k people
Uber does a lot more than that! If you want an honest accounting of what Uber's employees are doing you need to start with an honest listing of everything Uber actually does.
Could Uber be built with a lot less now days? Absolutely, but it's also an interesting balance of even if they could, if their cloud bill for X service is 10m a year, and they think their own team of 4 could build X internally for less then that...why not do that?
There's very little I can think of that seems to have changed about the site in the last 2 years for results from tripling their headcount.
Citation for the old headcount: https://www.theverge.com/2021/2/9/22274077/reddit-funding-ro...
Granted, I have no idea what scalability Reddit has done in 2 years. You can even argue it didn't need most of it (some, sure. Pandemic did surge traffic).
Under that you are going to have an advertising-related departments both in sales and engineering. Much of the custom engineering focus is going to be directed at ads and how to make money which users will never see. Remember that users are the product, not the customer.
Then there's going to be the forum site itself, and the problem of moderating the moderators via the admins and generating visibility into how mods and users are abusing the system are going to be larger than expected at their scale.
Then there's the problem of trying to keep up with automated bots and straight forwards abuse of the system and that arms race against other people's code.
Then reddit has to interface with government agencies and is now responsible for policing itself when it comes to foreign troll farms. There is likely groups of engineers and managers assigned to dealing with the government.
And there's just the complexities of being such a large global business, and needing to deal with worldwide regulations and being such a large regulatory target, which will complicate everything.
You can try to wave all that away and say that they should just embrace radical free speech and fire 75% of the people, but that would result in the site being taken over by Troll farms, Nazis, and CSAM -- with advertisers fleeing and the US government itching to regulate them.
(Although I'm sure they could fire the useless half of middle management that I'm certain that they've accumulated, but that wouldn't make that enormous of a dent in the headcount)
In general this is an xkcd/793 problem:
Only replace physicists with software engineers. It seems simple because you wave all the concerns you don't care about and you're entirely focused on just building a forum website for users. Without advertisers, bots, bad actors, worldwide regulations, meta-moderation, etc the problem dramatically simplifies itself, but that isn't actually what reddit does.
The first entry is "Nacho Fries are back for a limited time only" which I feel is the perfect answer to that question.
It's all too accurate.
Let's add up a few factors - Reddit is an 18yr old startup with Peter Pan syndrome. - Mountains of tech debt, like every other company - Industry incentives promote people to squirt out a feature, receive titles or money, and move on - Freedom of speech brings all the evils with it, like CP, that Reddit has complicated systems to catch and moderate - IPO desires mean financial belt tightening to look as mature as possible - Industry wide layoffs have created a fertile ground of new startups germinating. VCs are actively hunting for the next big AI unicorn. Torrential increases in traffic are hitting the company. And why? Reddit is a curation engine. The data is sorted and labeled to a degree. All this to say, there's a lot of refactoring needed on top of shifting ground created by the intersection of tech debt and IPO aspirations. - Guessing by all public measures, Reddit has a ton of traffic already. Extrapolating from other large companies I have worked for, the infrastructure and transport costs are significant. - 3P apps scrubbing Reddit ads and substituting their own are a brand and liability risk
TL;DR toil from legacy, more complexity than you can imagine, pressure of market
That said 2k people is a LOT, and it's not like the site itself is very high-tech or really pushing things forward. It's a aggregator of content with comments...nothing really advanced.
Yeah, 2k people is overkill.
Reddit should be ran more like HN or even craigslist. Ppl never wanted a js heavy UI change. they liked the bare bones old school aesthetic. It gave that feeling of nostalgia, one of the last early web apps.
If Reddit never did a redesign, never hired more staff than really need, they'd definitely be profitable by now.
sctb was previously on the mod team, but left YC a few years ago. He's recently reappeared in comments though.
Maybe two dozen engineers/devops, a dozen support/management people for them, and 50-100 minimum wage-ish admins and their managers/corporate support structure. I can't pretend to know anything about the ad sales part of the business, I guess that's where the other 1800 people live?
Narrowing focus to just engineering: how would you allocate 24 engineers across infra (networking, storage, scaling), backend (API, features), web frontend, design, mobile (iOS & Android), security, ads & monetization, Abuse & internal tooling dev? Or would you allocate 1 or 2 individuals to each (working 60-80 hour weeks) rather than teams (3-5 people, minimum) + line-managers & product folk. Add QA and you realize "2 dozen engineers" isn't a lot, the devil is in the details.
I was being charitable on 3-member teams: noone wants to be on an on-call roster that's only 3-engineers deep. Each of the areas I highlighted likely has multiple teams rather than just one (e.g. infra can be split into parthensized teams - networking, storage, operations/ops), and I intentionally completely ignored R&D functions.
Perhaps the market rewards growth more than it does efficiency? I can't name a single company (post WhatsApp) that has users in the 9-plus-digit-range and has less than a thousand employees.
It seems like investors prefer to have a large (but still profitable) org with a lot of redundancy and clear systems in place than an elite team of superheroes
I'm just annoyed that Spez is running a bloated overweight organization, then using that fact to justify some user-hostile money grubbing
I dislike unnecessary bloat, be it in software or in org design, as much as the next guy, but staking your company's long-term future on superheroes (or superpowers, as it's for some inexplicable reason still fashionable to say) is a recipe for disaster.
There are dozens of unprofitable Internet search engines each made by a handful of people, yet Google somehow employs ~200k people to run a profitable search engine.
I can create (probably build from scratch) a wiki in same amount of time but non-profit Wikipedia is 450(?) people.
The core tech isn’t the driver to headcount, the business model is.
For Reddit, that probably means selling ads and acquiring users. Which probably means a whole lot of ad tech and other optimizations/features to increase visitors to your site. Ad sales, ad tech, etc etc. It probably means ad sales to big clients with complex legal, etc.
In the tech, entire feature groups worth having because it serves important business goals around user retention or growth. Some small percentage of extra eyeballs helps Reddit towards these goals that a site like hacker news wouldn’t care about.
If anything, the fact that they have only 2k employees shows how much free labor the mods provide.
Twitter pre-pandemic: 5000
Wikipedia pre-pandemic: 450
So... maybe 1000.