I think this is a pretty naive take. Developing anything at Twitter-scale will run into security considerations, infrastructure development or optimization, constructing data workflows, multiple design iterations, UX design (how do people find and use this feature?), i18n, accessibility, product marketing, user testing, copy testing, and other functions and that’s not even considering the actual product development, which is of course across multiple platforms. See also: https://danluu.com/sounds-easy/

Instagram scaled to a 1bn usd valuation with 13 employees. Then they got acquired by facebook.

Instagram is still pretty lean from what I’ve heard.

Build the product, sure. Build the product to scale across the world?...I strongly disagree.

Pre-revenue that is possible. Post revenue, you'd need at least lawyers and accountants in each country you operate in. Then you need systems to respond to subpeonas and legal requests, that is a hundred people right there.

IG was 13 employees when you had a fraction of the images and they didnt have to be profitable, rather just grow by burning $. Now consider scale and profitability and you need even more people to grow tech cost consciously.

> Now consider scale and profitability and you need even more people to grow tech cost consciously.

This seems intuitively false. Adding more people to tech domains can actually decrease effectiveness with more expense, resulting in lower profits. Obviously, there is a sweet spot, but Twitter is a basic product regardless of how many countries it operates within.

I'm curious about your opinion on this. I may be thinking about this incorrectly.

Suppose you are Twitter and operating in, say, {Ghana, Russia, Korea, Mexico, and Pakistan}. You're served with subpoenas and legal IP requests in each country. If you dont respond in time, your license to operate may be revoked. Each request is in the native language. How do you "scale" this technically? Can one lawyer handle these across the world? How?

First recognize that lawyers are not a technical issue and that mostly they handle independent cases and subpoena requests. Since the work is fairly compartmentalized (unlike software development) throwing more people at the problem has less of a consequence. Hire a law firm in each country to handle your cases; No need to have lawyers all over the world as employees. Have a small team of international lawyers as employees or hire them as contractors to oversee the contracted teams of lawyers around the world and negotiate when needed.

Furthermore, developing tech for managing this is likely to be unnecessary as there are existing software packages already out there. Doing business across the world and handling legal issues is not unique to Twitter.

>> Hire a law firm in each country to handle your cases; No need to have lawyers all over the world as employees. Have a small team of international lawyers as employees or hire them as contractors to oversee the contracted teams of lawyers around the world and negotiate when needed.

I generally agree with your response, but note that you've just converted FTEs to contractors/outsourcers and "reduced" FTEs that way. You'd have to do the same with accountants and many other country-level positions. But w2-->1099 isnt really a "win".

I don't think Instagram had to seriously tackle the kinds of problems you're replying to prior to that acquisition. But when you're a big time tech company, the minimum bar is simply higher. That means more humans.

There is a huge difference between building a clone today on top of today's tools and commodity cloud platforms versus building it originally on what was around in 2006 and then having to constantly add features, change things, and then maintain 15 years worth of accumulated special edge cases forever without downtime.

Comparing to Instagram seems pretty unfair too as it's newer, it was even-more-featured-limited for a long time, is now part of a much larger behemoth and had much lower scale than today's twitter when it was acquired, and my guess would be that it's an easier scaling problem too (lower on post frequency, more read-heavy workload).

Speaking out of what experience? You've just appealed to your own authority; it seems not out of line to ask whence that comes.

Working at several big tech companies in industry. I can't say my experience is fully representative but I do start to see a pattern when my experiences line up with that of all my friends. There are a _lot_ of internal tools and anecdotally many of them seem like they're designed to abstract away things which wouldn't need abstracting for a company that exclusively hired high performing engineers.

Plenty of things are abstracted not because they need to be, but because they can be, and doing so increases efficiency, security, the ability to scale, and the ability of engineers to reason about the development.

And do the task consistently and correctly.

Even highly effective engineers will have trouble delivering features, in a large company, when they could easily deliver the same features in a smaller company. Process and politics is part of it, but the one of the problems with having so many developers, is that these developers end up writing code, all of that code ends up making things more complicated than they need to be, which makes introducing new features more complicated, due to all the systems you need to integrate with.