4 ms·

1.5 engineers?

nudpiedo 5y ago
A Person is working only 50% in That Project/Area due other responsibilities (project management, partial time, shared time with another team or whatever)
- ausudhz 5y ago
  Yes but doesn't work that way, with context switching cost and all of the rest people who are 50% in one projects are not really that effective. If anything, shows that they were not that committed to even put a resource dedicated to that project
  - tialaramex 5y ago
    It depends how granular the 50% is. I don't feel much cost from "context switching" overnight if Tuesday and Wednesday are project A time while Monday and Friday are project B time, but it gets to be a problem if I'm context switching hour by hour or taking random calls about A versus B during my normal working day.
    - ausudhz 5y ago
      Don't know where you work, but you're not switching from cooking to washing dishes. So yes there's a cost involved and is never a clear cut especially if you run projects in parallel.
      - adjkant 5y ago
        As someone who regularly works on a lot of projects at once or regularly context switching, I think this is missing the larger point here. Returns are not magically nothing for someone 50/50 on two projects, but there is of course a cost, and that cost increases with the number of projects, the complexity of the projects, and the pressure for delivery of the projects. It also varies person to person. Going back to your original post: > Yes but doesn't work that way, with context switching cost and all of the rest people who are 50% in one projects are not really that effective. Again, person dependent but missing the point. Would have it been more accurate to say they have 1.4327 engineers on the project? Probably. Is that helpful when everyone understands what 1.5 actually means? Not really. And two projects is not so unreasonable depending on the situation. > If anything, shows that they were not that committed to even put a resource dedicated to that project 100% yes to that. And as we have learned since this comment thread, it was not a 50% time situation but actually someone wearing two hats (PM and engineer): https://news.ycombinator.com/item?id=28426523 https://news.ycombinator.com/item?id=28426523
        ausudhz 5y ago
        > Again, person dependent but missing the point. Would have it been more accurate to say they have 1.4327 engineers on the project? You're reading too far, this is not the point of my comment? As you mentioned on the part below I was highlighting that they didn't want to put dedicated resource on the projects. The rest is just your own imagination ramping up to assume that I meant something else. And yes, we didn't know before that was one engineer doing also PM work. But so said, would've been more effective to have an engineer doing the work and a PM on two projects. Is easier to switch between managing stuff than engineering stuff.
        adjkant 5y ago
        > The rest is just your own imagination ramping up to assume that I meant something else. I mean it was a direct claim in your comment I was responding to, literally quoted which is the claim that someone on two project is "not really that effective". I'm simply offering my experiences that don't always match that. That's not my imagination there. No one here was arguing about that they clearly did not resource it in a meaningful way (I said as much in my post), that's not what this thread is discussing though...
- pcurve 5y ago
  I could be wrong but I took it as commentary on seemingly small number of headcount. (Vs fractional nature)
pcurve 5y ago
Good old days of development.
iamshs 5y ago
Full time equivalent most probably.
chubot 5y ago
The part that's more interesting to me is the "UI/Java" part. It's very hard to think of any consumer web company from that era that iterated on a UI quickly with Java and hit exponential growth. Here is a list of YC startups, and you have to scroll down before you hit anything but Ruby/Python (and C++ for back ends, presumably Go too). But I'd say this is true of all web companies in general, not just YC companies. (You'd probably see more PHP in non-YC companies, and I'd argue it's still more appropriate than Java for this use case) https://charliereese.ca/article/top-50-y-combinator-tech-startups https://charliereese.ca/article/top-50-y-combinator-tech-sta... Aside from the outliers that were GMail and Maps (acquisition), Google had a lot of trouble producing good UIs. Google+ was a good example of UI well below the state of the art at the time. Notably YouTube was a huge Python codebase ... which Google was already using at the time, but not much for web front ends.
- zorr 5y ago
  This email seems to be about 6 months before they released GWT 1.0. Their toolkit to compile Java to JS for web UI's. They likely already used it internally by then. If I remember correctly at the time they also used lots of Java on the backend.
- cmrdporcupine 5y ago
  I suspect what you're seeing is a bit of "I've got a hammer and everything looks like a nail" in relation to the fancy shiny GWT thing that some teams at Google built at the time. IMHO GWT was an awful toolkit and had a slow and terrible development cycle, but there was a lot of hype about it back then as if it was solving the bridge between dynamic front ends and backends written in Java. I wasn't even at Google at the time but had management types trying to foist it on me at a company I worked at, and it was frankly a failure. I'm not sure if Google Video was done in GWT, but it certainly had the look of a GWT product. When I came to Google in 2012 there were still some products using it but it was mostly recognized for being clunky and those products have been migrated.
  - nostrademons 5y ago
    Pretty sure it was GWT. I joined in 2009 when they were in the midst of the rewrite.
sulam 5y ago
Google internally has a kind of cultural belief that the company’s engineers are far more capable than most engineers in the rest of the world. It’s fairly arrogant, but some of it is founded on some real advantages around tooling and systems engineering.
- magicalist 5y ago
  Youtube was like 10 people at the time, so it's not ridiculous given they had the rest of Google's infrastructure to fall back on. Still wasn't sufficient, though :)
- sangnoir 5y ago
  IIRC, in that time-frame, Google had (one of) the highest revenue per engineer in the world- arrogant or not, they had good reasons for their self-belief.
  - sulam 5y ago
    So did SGI.
w0mbat 5y ago
I was the one front-end Java UI engineer and Nikhil (mentioned in the email) was the half. He was a PM who helped out on coding half-time. The Google Video team was very small.
- jtsiskin 5y ago
  What does you think of Peter Chane?
  - w0mbat 5y ago
    Nice guy, knows his stuff.

1.5 engineers?

A Person is working only 50% in That Project/Area due other responsibilities (project management, partial time, shared time with another team or whatever)

ausudhz 5y ago
Yes but doesn't work that way, with context switching cost and all of the rest people who are 50% in one projects are not really that effective. If anything, shows that they were not that committed to even put a resource dedicated to that project
- tialaramex 5y ago
  It depends how granular the 50% is. I don't feel much cost from "context switching" overnight if Tuesday and Wednesday are project A time while Monday and Friday are project B time, but it gets to be a problem if I'm context switching hour by hour or taking random calls about A versus B during my normal working day.
  - ausudhz 5y ago
    Don't know where you work, but you're not switching from cooking to washing dishes. So yes there's a cost involved and is never a clear cut especially if you run projects in parallel.
    - adjkant 5y ago
      As someone who regularly works on a lot of projects at once or regularly context switching, I think this is missing the larger point here. Returns are not magically nothing for someone 50/50 on two projects, but there is of course a cost, and that cost increases with the number of projects, the complexity of the projects, and the pressure for delivery of the projects. It also varies person to person. Going back to your original post: > Yes but doesn't work that way, with context switching cost and all of the rest people who are 50% in one projects are not really that effective. Again, person dependent but missing the point. Would have it been more accurate to say they have 1.4327 engineers on the project? Probably. Is that helpful when everyone understands what 1.5 actually means? Not really. And two projects is not so unreasonable depending on the situation. > If anything, shows that they were not that committed to even put a resource dedicated to that project 100% yes to that. And as we have learned since this comment thread, it was not a 50% time situation but actually someone wearing two hats (PM and engineer): https://news.ycombinator.com/item?id=28426523 https://news.ycombinator.com/item?id=28426523
      - ausudhz 5y ago
        > Again, person dependent but missing the point. Would have it been more accurate to say they have 1.4327 engineers on the project? You're reading too far, this is not the point of my comment? As you mentioned on the part below I was highlighting that they didn't want to put dedicated resource on the projects. The rest is just your own imagination ramping up to assume that I meant something else. And yes, we didn't know before that was one engineer doing also PM work. But so said, would've been more effective to have an engineer doing the work and a PM on two projects. Is easier to switch between managing stuff than engineering stuff.
        adjkant 5y ago
        > The rest is just your own imagination ramping up to assume that I meant something else. I mean it was a direct claim in your comment I was responding to, literally quoted which is the claim that someone on two project is "not really that effective". I'm simply offering my experiences that don't always match that. That's not my imagination there. No one here was arguing about that they clearly did not resource it in a meaningful way (I said as much in my post), that's not what this thread is discussing though...
pcurve 5y ago
I could be wrong but I took it as commentary on seemingly small number of headcount. (Vs fractional nature)

pcurve 5y ago

Good old days of development.

iamshs 5y ago

Full time equivalent most probably.

chubot 5y ago

The part that's more interesting to me is the "UI/Java" part. It's very hard to think of any consumer web company from that era that iterated on a UI quickly with Java and hit exponential growth. Here is a list of YC startups, and you have to scroll down before you hit anything but Ruby/Python (and C++ for back ends, presumably Go too). But I'd say this is true of all web companies in general, not just YC companies. (You'd probably see more PHP in non-YC companies, and I'd argue it's still more appropriate than Java for this use case) https://charliereese.ca/article/top-50-y-combinator-tech-startups https://charliereese.ca/article/top-50-y-combinator-tech-sta... Aside from the outliers that were GMail and Maps (acquisition), Google had a lot of trouble producing good UIs. Google+ was a good example of UI well below the state of the art at the time. Notably YouTube was a huge Python codebase ... which Google was already using at the time, but not much for web front ends.

zorr 5y ago
This email seems to be about 6 months before they released GWT 1.0. Their toolkit to compile Java to JS for web UI's. They likely already used it internally by then. If I remember correctly at the time they also used lots of Java on the backend.
cmrdporcupine 5y ago
I suspect what you're seeing is a bit of "I've got a hammer and everything looks like a nail" in relation to the fancy shiny GWT thing that some teams at Google built at the time. IMHO GWT was an awful toolkit and had a slow and terrible development cycle, but there was a lot of hype about it back then as if it was solving the bridge between dynamic front ends and backends written in Java. I wasn't even at Google at the time but had management types trying to foist it on me at a company I worked at, and it was frankly a failure. I'm not sure if Google Video was done in GWT, but it certainly had the look of a GWT product. When I came to Google in 2012 there were still some products using it but it was mostly recognized for being clunky and those products have been migrated.
- nostrademons 5y ago
  Pretty sure it was GWT. I joined in 2009 when they were in the midst of the rewrite.

sulam 5y ago

Google internally has a kind of cultural belief that the company’s engineers are far more capable than most engineers in the rest of the world. It’s fairly arrogant, but some of it is founded on some real advantages around tooling and systems engineering.

magicalist 5y ago
Youtube was like 10 people at the time, so it's not ridiculous given they had the rest of Google's infrastructure to fall back on. Still wasn't sufficient, though :)
sangnoir 5y ago
IIRC, in that time-frame, Google had (one of) the highest revenue per engineer in the world- arrogant or not, they had good reasons for their self-belief.
- sulam 5y ago
  So did SGI.

w0mbat 5y ago

I was the one front-end Java UI engineer and Nikhil (mentioned in the email) was the half. He was a PM who helped out on coding half-time. The Google Video team was very small.

jtsiskin 5y ago
What does you think of Peter Chane?
- w0mbat 5y ago
  Nice guy, knows his stuff.