4 ms·

Those things you mention hardly justify thousands of expensive developers to maintain and create. I bet a team of 10 top notch engineers could create Spaces+Blu

Those things you mention hardly justify thousands of expensive developers to maintain and create. I bet a team of 10 top notch engineers could create Spaces+Blue+Crypto pfp's in a couple months and run it at Twitter scale. Don't forget Instagram was acquired when it had 13 employees serving tens of millions of users. Speaking out of experience, most engineers in big tech are bike shedding on internal tools that don't do anything useful. A small minority deliver the majority of the impact. On top of that at companies like Twitter and Google some of those useless employees spend their time complaining about social justice initiatives rather than doing work.

throwanem 5y ago
Speaking out of what experience? You've just appealed to your own authority; it seems not out of line to ask whence that comes.
- ceeplusplus 5y ago
  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.
  - WesternWind 5y ago
    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.
    - djbusby 5y ago
      And do the task consistently and correctly.
dkislyuk 5y ago
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/ https://danluu.com/sounds-easy/
- mccorrinall 5y ago
  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.
  - phillipcarter 5y ago
    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.
  - TuringNYC 5y ago
    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.
    - cudgy 4y ago
      > 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.
WatchDog 5y ago
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.
majormajor 5y ago
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).

throwanem 5y ago

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

ceeplusplus 5y ago
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.
- WesternWind 5y ago
  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.
  - djbusby 5y ago
    And do the task consistently and correctly.

dkislyuk 5y ago

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/ https://danluu.com/sounds-easy/

mccorrinall 5y ago
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.
- phillipcarter 5y ago
  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.
- TuringNYC 5y ago
  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.
  - cudgy 4y ago
    > 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.

WatchDog 5y ago

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.

majormajor 5y ago

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).