4 ms·

That means it had revenue that was growing. We don't know what the expenses were. It's still possible it wasn't profitable.

koolba 3y ago
We’re taking about a database that stores on the order of 1KB per customer. The product operating costs round down to zero per customer. The only real expenses would be dealing with customer service and card fraud. While it’s possible that those costs could be substantial, Google isn’t exactly known for throwing a lot of money at customer service.
- dijit 3y ago
  Internally; I believe google is super averse to anything that scales linearly or faster than linearly. - at least as a rule. So it doesnt matter, i speculate, if it was earning $1M in profit a year for 10M domains if they need to employ 50 call center staff - and for the next 10M domains they need to hire another 50 call center staff— I think they would rather pull the plug. Which I personally would never do, but, it is in line with other decisions they’ve made and how I have heard key people talk about googles leadership externally.
  - deelowe 3y ago
    > Internally; I believe google is super averse to anything that scales linearly or faster than linearly. - at least as a rule. This is 1000% true. The culture is that diminishing returns means linear scale will eventually turn into a cost center. Execs only care about 10x initiatives. My understanding is that Larry personally canceled gFiber b/c the margins, while profitable, were too thin.
    - lesuorac 3y ago
      I suspect even if gFiber had super high margins it would've still had its operations scaled down. It's a huge pain in the ass to deploy fiber. The cities that got Google Fiber mostly had to bend over backwards in order to do so. There's a lot of logistical (and legal) factors that just don't exist for pretty much any other Google product. Like what approvals does Google need to release Bard? But you could make pages of approvals needed to deploy fiber to a city.
      - unilynx 3y ago
        > Like what approvals does Google need to release Bard Well, for one they still need Ireland's DPA approval before they release it to the EU.
        lesuorac 3y ago
        Surely if Google Fiber started EU operations they would also need DPA approval? The point is that fiber would require a lot more approvals than many of the pure software projects. Also afaik, fiber never launched in EU and exists in many less locations than Bard.
    - tyingq 3y ago
      >My understanding is that Larry personally canceled gFiber b/c the margins, while profitable, were too thin. That makes sense given "nothing linear", but I can't understand why they would have thought it could have been better. Same for domains. In what scenario would they grow exponentially?
      - dijit 3y ago
        if they can automate the support fully? like someone else in this thread said; in theory it’s a global distribution of 1kb in a database and a user interface. if you are the best at AI, maybe you can solve the support burden. Or at least thats what they attempted.
    - thereisnospork 3y ago
      It feels like such a myopic decision on several fronts, not least of all the obvious cross-product synergies. Google Fiber customers should have been the target market for Stadia, should have been a backbone for deploying semi-public wifi for Pixel Phones owners, a target audience for 'cinephile-grade' upgraded youtube subscription, etc. Nevermind that non-linearity should have been achievable[0] via lobbying at the state/federal level. [0]An advantage only really achievable by a company as large as google, with money and weight to throw around.
- VikingCoder 3y ago
  Sure but the database they used probably had seventy-four nines of uptime and used atomic clocks to support nearly simultaneous changes to a single record at a rate of a hundred and six thousand hertz, and that thing was crazy expensive. Unfortunately that database was built on Awesome Google Internal Technology 884611, which has been deprecated, and the replacement system doesn't work yet. The team that built Domains ended up owning Awesome Google Internal Technology 884611 because they were supposedly the only team left still using it, but then they discovered that Google Search also relies on it through a Stubby that no one noticed. So then the Domains oncalls were getting swamped with bug reports from Google Search every time someone used Google Image Search in Italy, for some reason... The Domains SWEs and SREs put up a heroic effort to keep Search (and Domains) running, but it was noted that implementing something like Domains would neither be Difficult or Impactful, so they all got Needs Improvement and resigned to go work on something more rewarding like High Frequency Trading of Kale Futures or driving a Lyft. Google Image Search in Italy has been flaky ever since.
kllrnohj 3y ago
> It's still possible it wasn't profitable. Surely that's a lot more trivially fixed with a small price bump than throwing the entire revenue stream & user sentiment into the trash? Google Domains is currently charging $12/year for a .com domain, which we know from cloudflare the "at cost" price of that is $9.15/year. Comparing the $12/year to other providers it's kinda middle-of-the-road, not especially cheap or anything. Shopify is $15/year while porkbun is ~$10/year, so $12/year certainly appears to be a price that has adequate margin to either be profitable or if not only needs a small bump to be profitable?

koolba 3y ago

We’re taking about a database that stores on the order of 1KB per customer. The product operating costs round down to zero per customer. The only real expenses would be dealing with customer service and card fraud. While it’s possible that those costs could be substantial, Google isn’t exactly known for throwing a lot of money at customer service.

dijit 3y ago
Internally; I believe google is super averse to anything that scales linearly or faster than linearly. - at least as a rule. So it doesnt matter, i speculate, if it was earning $1M in profit a year for 10M domains if they need to employ 50 call center staff - and for the next 10M domains they need to hire another 50 call center staff— I think they would rather pull the plug. Which I personally would never do, but, it is in line with other decisions they’ve made and how I have heard key people talk about googles leadership externally.
- deelowe 3y ago
  > Internally; I believe google is super averse to anything that scales linearly or faster than linearly. - at least as a rule. This is 1000% true. The culture is that diminishing returns means linear scale will eventually turn into a cost center. Execs only care about 10x initiatives. My understanding is that Larry personally canceled gFiber b/c the margins, while profitable, were too thin.
  - lesuorac 3y ago
    I suspect even if gFiber had super high margins it would've still had its operations scaled down. It's a huge pain in the ass to deploy fiber. The cities that got Google Fiber mostly had to bend over backwards in order to do so. There's a lot of logistical (and legal) factors that just don't exist for pretty much any other Google product. Like what approvals does Google need to release Bard? But you could make pages of approvals needed to deploy fiber to a city.
    - unilynx 3y ago
      > Like what approvals does Google need to release Bard Well, for one they still need Ireland's DPA approval before they release it to the EU.
      - lesuorac 3y ago
        Surely if Google Fiber started EU operations they would also need DPA approval? The point is that fiber would require a lot more approvals than many of the pure software projects. Also afaik, fiber never launched in EU and exists in many less locations than Bard.
  - tyingq 3y ago
    >My understanding is that Larry personally canceled gFiber b/c the margins, while profitable, were too thin. That makes sense given "nothing linear", but I can't understand why they would have thought it could have been better. Same for domains. In what scenario would they grow exponentially?
    - dijit 3y ago
      if they can automate the support fully? like someone else in this thread said; in theory it’s a global distribution of 1kb in a database and a user interface. if you are the best at AI, maybe you can solve the support burden. Or at least thats what they attempted.
  - thereisnospork 3y ago
    It feels like such a myopic decision on several fronts, not least of all the obvious cross-product synergies. Google Fiber customers should have been the target market for Stadia, should have been a backbone for deploying semi-public wifi for Pixel Phones owners, a target audience for 'cinephile-grade' upgraded youtube subscription, etc. Nevermind that non-linearity should have been achievable[0] via lobbying at the state/federal level. [0]An advantage only really achievable by a company as large as google, with money and weight to throw around.
VikingCoder 3y ago
Sure but the database they used probably had seventy-four nines of uptime and used atomic clocks to support nearly simultaneous changes to a single record at a rate of a hundred and six thousand hertz, and that thing was crazy expensive. Unfortunately that database was built on Awesome Google Internal Technology 884611, which has been deprecated, and the replacement system doesn't work yet. The team that built Domains ended up owning Awesome Google Internal Technology 884611 because they were supposedly the only team left still using it, but then they discovered that Google Search also relies on it through a Stubby that no one noticed. So then the Domains oncalls were getting swamped with bug reports from Google Search every time someone used Google Image Search in Italy, for some reason... The Domains SWEs and SREs put up a heroic effort to keep Search (and Domains) running, but it was noted that implementing something like Domains would neither be Difficult or Impactful, so they all got Needs Improvement and resigned to go work on something more rewarding like High Frequency Trading of Kale Futures or driving a Lyft. Google Image Search in Italy has been flaky ever since.

kllrnohj 3y ago

> It's still possible it wasn't profitable. Surely that's a lot more trivially fixed with a small price bump than throwing the entire revenue stream & user sentiment into the trash? Google Domains is currently charging $12/year for a .com domain, which we know from cloudflare the "at cost" price of that is $9.15/year. Comparing the $12/year to other providers it's kinda middle-of-the-road, not especially cheap or anything. Shopify is $15/year while porkbun is ~$10/year, so $12/year certainly appears to be a price that has adequate margin to either be profitable or if not only needs a small bump to be profitable?