4 ms·

Whenever someone says "ops doesn't matter" I'm reminded of Friendster's huge early-advantage lead that was squandered because they couldn't keep the site up for

by alrs 3y ago

Whenever someone says "ops doesn't matter" I'm reminded of Friendster's huge early-advantage lead that was squandered because they couldn't keep the site up for more than six hours a day.

colesantiago 3y ago
Surely it must be easier this time to build something like Facebook today right?
- saulpw 3y ago
  easier != easy
- danielhlockard 3y ago
  I can't imagine it wouldn't be easier to build something like early FaceBook today. There was no GCP / AWS / Azure / whatever back then. We were still racking servers, and we liked it.
  - beshrkayali 3y ago
    If you consider cost associated as an added difficulty, it’s possibly even more difficult now.
    - fastball 3y ago
      What has gone up in cost? The methods available 20 years ago are still available now. We just have additional options now, which are actually cheaper to start, as you can get a lot of usage out of AWS' free tier. Sure, once you get to "scale up" it's probably cheaper to do everything on-prem still, but that story is much better too with things like Golang, Docker, Elixir, k8s, etc, etc to make managing/building web-scale infrastructure much easier.
      - saulpw 3y ago
        There were 1 billion people on the internet in 2005. Now there are more than 5 billion people on the internet. There were no smartphones in 2005 and everyone used the web to access "social media". Now any social media is nonviable unless it has an app for both of the smartphone platforms. Youtube and streaming video didn't exist in 2005. Now it's table stakes for social media to allow people to share videos. The bar has been raised by at least an order of magnitude. Sure, it might be easier and/or cheaper now to do what Facebook did...in 2005. But that won't get you anywhere in 2023.
  - ddorian43 3y ago
    There are no more vc funds (currently) to pay your cloud bill though.
    - vasco 3y ago
      Did you add the words "generative AI" to your pitch?
  - flashback2199 3y ago
    > I can't imagine it wouldn't be Whenever you have a double negative, flip both: "I can imagine it would be" If that changes the meaning too much, you can say: "I can imagine it might be"
    - a1369209993 3y ago
      > > I can't imagine it wouldn't be > I can imagine it would be No. That means something completely different. They're saying (probably hyperbolically) that they cannot imagine the first of two possible situations, not that they can imagine the second.
      - flashback2199 3y ago
        Oh don't worry, I understand that just fine. It's the speaker's or writer's task to iterate for a half second to find whatever non double negative sentence fits what they want to say. English isn't code. You can't say double negatives in conversation or writing, and you have to find the closest thing without using a double negative.
        a1369209993 3y ago
        > You can't say double negatives It is in fact not the case that you can't say that. Double negatives, like sentences with a excessive number of conjuctions, are perfectly grammatical (when they don't violate some actual grammatical rule like the one "You do can say that." does); rather, they are in most cases stylistically bad, for much the same reasons that: stop = 0 while(a() && !stop): b() if(!(c==d || !e())) stop = 1 // should be `if(c!=d && e()) break` if(!stop) e() is stylistically bad, namely excessive and more importantly needless complexity.
        flashback2199 3y ago
        Thanks, we agree.
        3y ago
- bdcravens 3y ago
  The application? Sure. Most of the functionality could be hacked out in a "weekend" with a framework like Rails or Laravel. The network? Not really. As a matter of fact, I don't think a new perfect FB would even be used today; people are only there because the people they want to reach already are. People today want a faster social experience (a la TikTok)
  - spike021 3y ago
    >The network? Not really. As a matter of fact, I don't think a new perfect FB would even be used today; people are only there because the people they want to reach already are. Good example for this is Threads. Everybody I know and care to follow are still on Twitter. Threads was cool the first week or so but after that Twitter was too strong of a magnet. I don't know how a new network gets built these days simply because the major ones that exist are already so engrained. Politics and whatnot of Twitter aside, of course.
    - trilbyglens 3y ago
      Special interest networks still have a big advantage over the giant general ones.
- oooyay 3y ago
  Are you talking about building a Facebook clone of just the website and API? Yeah, probably relatively easy. If you're talking about data warehousing, research, and building/tweaking their algorithms, no that's still hard.
cameldrv 3y ago
It wasn't just ops though, it was also their product strategy, algorithms, and scaling approach. One of the features of Friendster was that when you viewed someone's profile, it would show you all of the ways you were connected to that person, out to I think four degrees. I'm not sure how they did this, but it didn't scale, and the result was that in I believe late 2003, you'd see everyone in a coffee shop with their laptops out waiting for a friendster page to load. It would take over ten minutes to load a page sometimes. The next year, MySpace started getting popular. MySpace didn't do all of this complicated social graph calculation at all. It would just tell you if you were friends or not. Naturally it was way faster, and despite other advantages of Friendster, it didn't take 10 minutes to load a page, and it took over completely until Facebook.
sib301 3y ago
I’ve never heard anyone say “ops doesn’t matter”
- ricc 3y ago
  I think it's usually implicitly communicated whenever there's a decision point to choose between allocating resources to ops or, for example, delivering a new feature. The new feature usually always wins.

by alrs 3y ago

Whenever someone says "ops doesn't matter" I'm reminded of Friendster's huge early-advantage lead that was squandered because they couldn't keep the site up for more than six hours a day.

colesantiago 3y ago

Surely it must be easier this time to build something like Facebook today right?

saulpw 3y ago
easier != easy
danielhlockard 3y ago
I can't imagine it wouldn't be easier to build something like early FaceBook today. There was no GCP / AWS / Azure / whatever back then. We were still racking servers, and we liked it.
- beshrkayali 3y ago
  If you consider cost associated as an added difficulty, it’s possibly even more difficult now.
  - fastball 3y ago
    What has gone up in cost? The methods available 20 years ago are still available now. We just have additional options now, which are actually cheaper to start, as you can get a lot of usage out of AWS' free tier. Sure, once you get to "scale up" it's probably cheaper to do everything on-prem still, but that story is much better too with things like Golang, Docker, Elixir, k8s, etc, etc to make managing/building web-scale infrastructure much easier.
    - saulpw 3y ago
      There were 1 billion people on the internet in 2005. Now there are more than 5 billion people on the internet. There were no smartphones in 2005 and everyone used the web to access "social media". Now any social media is nonviable unless it has an app for both of the smartphone platforms. Youtube and streaming video didn't exist in 2005. Now it's table stakes for social media to allow people to share videos. The bar has been raised by at least an order of magnitude. Sure, it might be easier and/or cheaper now to do what Facebook did...in 2005. But that won't get you anywhere in 2023.
- ddorian43 3y ago
  There are no more vc funds (currently) to pay your cloud bill though.
  - vasco 3y ago
    Did you add the words "generative AI" to your pitch?
- flashback2199 3y ago
  > I can't imagine it wouldn't be Whenever you have a double negative, flip both: "I can imagine it would be" If that changes the meaning too much, you can say: "I can imagine it might be"
  - a1369209993 3y ago
    > > I can't imagine it wouldn't be > I can imagine it would be No. That means something completely different. They're saying (probably hyperbolically) that they cannot imagine the first of two possible situations, not that they can imagine the second.
    - flashback2199 3y ago
      Oh don't worry, I understand that just fine. It's the speaker's or writer's task to iterate for a half second to find whatever non double negative sentence fits what they want to say. English isn't code. You can't say double negatives in conversation or writing, and you have to find the closest thing without using a double negative.
      - a1369209993 3y ago
        > You can't say double negatives It is in fact not the case that you can't say that. Double negatives, like sentences with a excessive number of conjuctions, are perfectly grammatical (when they don't violate some actual grammatical rule like the one "You do can say that." does); rather, they are in most cases stylistically bad, for much the same reasons that: stop = 0 while(a() && !stop): b() if(!(c==d || !e())) stop = 1 // should be `if(c!=d && e()) break` if(!stop) e() is stylistically bad, namely excessive and more importantly needless complexity.
        flashback2199 3y ago
        Thanks, we agree.
        3y ago
bdcravens 3y ago
The application? Sure. Most of the functionality could be hacked out in a "weekend" with a framework like Rails or Laravel. The network? Not really. As a matter of fact, I don't think a new perfect FB would even be used today; people are only there because the people they want to reach already are. People today want a faster social experience (a la TikTok)
- spike021 3y ago
  >The network? Not really. As a matter of fact, I don't think a new perfect FB would even be used today; people are only there because the people they want to reach already are. Good example for this is Threads. Everybody I know and care to follow are still on Twitter. Threads was cool the first week or so but after that Twitter was too strong of a magnet. I don't know how a new network gets built these days simply because the major ones that exist are already so engrained. Politics and whatnot of Twitter aside, of course.
  - trilbyglens 3y ago
    Special interest networks still have a big advantage over the giant general ones.
oooyay 3y ago
Are you talking about building a Facebook clone of just the website and API? Yeah, probably relatively easy. If you're talking about data warehousing, research, and building/tweaking their algorithms, no that's still hard.

cameldrv 3y ago

It wasn't just ops though, it was also their product strategy, algorithms, and scaling approach. One of the features of Friendster was that when you viewed someone's profile, it would show you all of the ways you were connected to that person, out to I think four degrees. I'm not sure how they did this, but it didn't scale, and the result was that in I believe late 2003, you'd see everyone in a coffee shop with their laptops out waiting for a friendster page to load. It would take over ten minutes to load a page sometimes. The next year, MySpace started getting popular. MySpace didn't do all of this complicated social graph calculation at all. It would just tell you if you were friends or not. Naturally it was way faster, and despite other advantages of Friendster, it didn't take 10 minutes to load a page, and it took over completely until Facebook.

sib301 3y ago

I’ve never heard anyone say “ops doesn’t matter”

ricc 3y ago
I think it's usually implicitly communicated whenever there's a decision point to choose between allocating resources to ops or, for example, delivering a new feature. The new feature usually always wins.