Hacker News Reader

As someone who joined pre-pandemic, my opinion is that we were unsustainably lean at the time. I was shocked at how small the infrastructure team was when I joined the company and things were honestly a bit rickety at the user scale we were at then. While engineering was probably the single biggest department, a large chunk of that 200 were in departments like Trust & Safety and Customer Experience (support).

We saw a huge amount of user growth during the pandemic, we massively increased the number of users that could be in a single guild (server), we launched features that asked a lot more from our AV backend (stage channels, camera video in guild voice), had to deal with a massive increase in spam during the crypto boom, expanded voice calls to Xbox and PS5 and had to deal with an increasingly fraught regulatory environment.

We also shipped a lot of user-facing features during that timeframe. Some of those ultimately got removed because of poor results and depending on your use of the platform I imagine some of those that stayed may not be particularly visible, but we really did ship a lot of stuff and I'm not going to attempt to list it all here.

Obviously, in hindsight we should have had a tighter focus and probably hired a bit less, but this sentiment that we are/were hugely bloated feels pretty wild to me from where I'm sitting.

This is great context and I really appreciate it. I honestly thought of Discord as being super stable even in 2020; interesting to hear things felt rickety, internally.

My sense is that Discord was exploring a lot of different ideas for monetization all at once, and that maybe the sprawl of new projects got unwieldy, esp. in 2022-23; how accurate would you say that is?

Still -- my quick google came up with 750 employees mid 2023. So this is 125-ish employees. Particularly with 5 months severance, that seems too small to save much money (assuming $300k fully loaded, that's $40m pa) and also too small to materially impact velocity by reducing headcount and being more careful about commits.

I'd expect more like a 50% cut to really speed up the significantly fewer projects by eliminating coordination overhead and giving PMs the ability to really slash projects.

ps -- I'm not claiming it was bloated, that's (implicitly, and I believe fairly?) the ceo's claim.

The article quotes the memo saying it is 170 employees.

It’s the way HN continues to rationalize the hundreds of thousands of layoffs. Every company is bloated and few software engineers are worth keeping it seems.

Knowing the average SV interview loop too well, I can't imagine how many bad (for corporate type work) hires growing that fast brings in. A bunch of leetcode-hard optimized engineers being thrown into an organization that just grew from 200 to 1000 people.

If you are part of that 200 original group, you probably have a paycheck for another 5 years just answering tech debt questions

When hiring fast, there's also an element of "we'll find something for you to do", meaning that you're expected to find something yourself. This can lead to people becoming very narrow on something that the business doesn't really value

Or better yet, "We hired you to do some work" but at performance review time "tell us how you have made an impact"

Dear company, I did what you asked me to do. If my work doesn't have adequate impact, that is on you for planning poorly.

thats the meaning of working for engineering-driven company. Engineers are trusted to collaborate among themselves and drive feature roadmap, and to make an impact.

Leadership doesnt trust non-engineers to handhold engineers and spoon feed feature requests, all non-engineer types should be focused on selling/marketing product or customer support

Brilliant! As if a solid feature roadmap doesn't require inputs from sales, marketing or customer support.

what makes you think that engineering-driven cultures ignore or don't ask for input from sales/marketing/customer support?

that's the collaboration part. But the final decision over prioritization of feature X over feature Y and planning is done by engineers, not salesmen

> what makes you think that engineering-driven cultures ignore or don't ask for input from sales/marketing/customer support?

The part where managers who don't partake in these efforts refuse to believe engineering estimates, engineering challenges and unknown unknowns and yet feel empowered to blame engineers in performance reviews.

That's all fine and dandy when you are small, and have a bunch of generalists that will throw themselves at any problem there is.

But when you grow to the point of having a sizeable middle management layer that isolates engineers from one another, by enforcing strict area of interest for each team, you really shouldn't call yourself "engineering driven" anymore. And also, the more you scale, the more specialists you need. Specialists that aren't able, or just aren't willing to move from place to place rapidly

Unless your client base is exclusively the same kind of engineer employed by said company, how could that possibly work well?

Goggle started it, I think it worked well for 1.9 trillion company as gogel

It sounds like a recipe for ruining a product and turning a company into a dysfunctional organisation. Onboarding 800 idle people prepared to reverse linked-lists :-)

Seems like it ends up with lot features that few people asked. And then another bunch that no one asked. And then outright user hostile things like reworking entire UI time and again and somehow it being worse each time...

Well maybe they just have like a lot of linked lists that they need reversed? With that many people working in parallel you could really increase the throughput of the reversing process. Of course this is assuming that you previously hired people to identify and remove any cyclical nodes in those link lists.

Tell me, what's the big O of having the people working in parallel on the reversing? And as a follow-up, can you explain whether it is better if you have a team of n people doing the cycle detection to do it using a system which is eventually consistent or strongly consistent and why?

Now imagine if they opened those jobs to all of us in low cost of living areas instead of demanding in person presence. They'd get better people, and people like me would be getting a decent bump even if we were only getting 50 or 60% of SF salary.

Why would they give you 50 or 60% of SF salary when they could hire someone in a real low cost area then?

Exactly - I feel that folks in the US or Western Europe who are in "low cost areas" are in for a shock, as in many ways they have the worst of all possible combos. Either you're in a big "tech city" where being present (often in person) provides a tangible benefit that your employer is willing to pay for, or your essentially bidding on cost, where someone in a lower-cost country will be able to beat you every time.

I've been in tech for a long time, but this is the first time we've had a significant downturn where remote/videoconferencing options have been so good. In the past it felt like executive cries of "we'll just outsource everything!" always backfired because the communication and management overhead (not to mention varying quality levels) had been extreme. Now, though, with everyone used to remote work, why bother looking to save a few bucks hiring in Wichita when you can save more, and get just as good if not better quality, hiring from Cordoba Argentina?

Because they already hired those people. There aren’t millions of super high skilled software engineers in Argentina or where ever, there are a limited quantity, and they are likely already employed somewhere.

Time Zone, English proficiency, familiar tax and legal regime and cultural affinity? When you get basically the same employee for 50% less, that's really appealing.

The people in the "real low cost area" aren't the same employee - I'm not saying they are less capable, but most companies are afraid of the significant hidden costs when trying to go too cheap this way.

Discord interview questions are not your typical leetcode problems. Your premise is wrong.

The graded lunch is dumb as fuck.

“Do you play any video games?”

“I’ve been playing a lot of KSP, but I’m not sold on KSP2 yet.”

“Yeah, I don’t play that game.” STRONG REJECT

Never heard of that happening, and neither would you, since you do not see interview feedback. Please don't make up a story.

Bwhahahahahahahahaha

Anonymous shill calls named lived experience a liar. What a time to be alive!

So you did somehow get access to your interview feedback? And you’re sure that your lunch buddy gave you a strong reject and swayed every other interviewer who otherwise gave you a positive review?

One of the most surprising things to me during the pandemic was the dramatic hiring spree that almost every major tech company engaged in at the same time. I was getting pings from different recruiters multiple times a month during this time, and I’m sure others were getting many more.

Today’s layoffs are a direct result of the irrational exuberance of that era.

Can someone explain this systemic hiring phenomena? A pandemic hits and every corporation thinks that's the time to hire and scale by gross multipliers? Why? Whose idea was this? How did it catch on everywhere?

The lockdowns which came about as a pandemic response resulted in a huge shift in spending - many companies were suddenly flush with cash, and many of them decided to use that cash to invest in their business, by hiring engineers.

With more engineers you could make more babies per month or something.

A lot of companies saw unexpected spikes in revenue and wanted to strike while the iron was hot. The impulse was rational but the scale of the response was completely out of proportion with reality. But it's not like we saw hiring sprees in airlines and hotel resorts. We saw sprees in industries which benefitted from people staying at home and communicating digitally (like Discord).

A lot of companies also saw spikes in revenue, raised money from investors at sweet terms based on that revenue growth (despite it being unsustainable), and then needed to spend that money to show they were going to use it to keep the good times rolling.

Stock market was climbing and everyone thought digital was going to be permanently the new default, so companies tried to move quickly to land grab all this potential market.

Federal pandemic money plus we were still at super-low interest rates.

Now the Federal spigot has dried up (well, is back to normal) and interest rates are higher.

Not every corporation, if I remember correctly Apple was quite restrained.

Investors want more growth and that translates to hiring more people to do more things.

Then it’s painful when more people slow things down because there wasn’t much process in place before the blitz hiring.

This isn’t a problem unique to engineering, you see it a lot with sales too. Leaders aren’t experienced or self reflected enough to work smarter.

> what do they even do all day?

Discord has added a bunch of new features over the past couple years. Unfortunately, they don't seem to be very useful features (at least not to me). I still just use it for voice chat during video games.

Their eng team wasn't particularly large. ~300 was what I last heard.

I don't know how eng grew over that time, but it doesn't strike me as bloated given their scale and user base.

> Clearly they are making non-trivial revenue

"Non-trivial" is a very low bar to clear. It's obvious that Discord simply does not need to be the scale it is. It doesn't need a billion dollars and it doesn't need a thousand employees. It's very clear that they only have the employees because the investors were stupid enough to give them that much money, and they need to find something to do with it.

Some products just do not need to be that big.

What on earth is your reasoning for saying this? I really dislike the habit of stating something absolutely ludicrous and running away with it.

Off the top of my head I cannot recall any big features added to discord over the last few years. I'd like to know what those 4x more employees were working on.

Maybe not engineers? Discord quickly morphed from a place where gamers yell at each other into a place where lots of communities and movements exist. Maybe that skyrocketed the need for moderation, lawsuits of people demanding that their former block party co-organizer gets un-admin'ed, abuse reports, etc?

They have been trying a lot of different routes for monetization, and over a pretty short time period. Adding lots of little things to Nitro, more community server monetization features; at one point, Discord was even trying to sell games a la Steam.

I'm sure that last one in particular had quite a lot of headcount associated, and it seems obvious that after they totally shuttered the project, they just reallocated the employees to new stuff.

They changed the icons for mute mic, and mute sound buttons. So that probably took at least a dedicated team and few months of work...

Activities and soundboards are the ones that come to mind. Activities are a pretty heavy feature.

It seems most, if not all, big-successful companies use the formula to over higher and then you prune off the lowest-poorest performers - however that metric is determined.

Discord is like nuclear fusion - always one year from being profitable.