4 ms·

Actually what happens in governments is that they sign a contract to provide product X for $Y. (i.e. Provide a new sidewalk for $100,000) Then through further

Actually what happens in governments is that they sign a contract to provide product X for $Y. (i.e. Provide a new sidewalk for $100,000) Then through further discussion, after the contract is signed, everyone realizes they actually need product X2. And so the contractor revises their bid to $Z where $Z exceeds the original $Y by a wide amount (i.e. Provide a new sidewalk and re-landscape the grass on either side for $250,000) There are no cost overruns in government contracts. Just changes in scope. You don't find out what they really need until you really get into talking with them.

jonknee 15y ago
> There are no cost overruns in government contracts. Just changes in scope. You don't find out what they really need until you really get into talking with them. Umm, there are tons of cost overruns in government contracts (and in non-government contracts). Consider the James Webb Space Telescope for example. Or the F-35. Or about any software project. It's very hard to predict costs of new inventions.
- angersock 15y ago
  So, in regards to fighterjets and things at least... I've heard (and I'd love to hear whether or not this sounds reasonable) that one of the big problems is that you'll source a part like a resistor, ASIC, or something, and then for whatever reason it'll get changed--maybe a cheaper part became available and you were forced to use the cheaper part, maybe the part is no longer in production and has to be switched out, maybe the part just failed to function correctly. The effect of this, of course, is that you have to redo all of your integration testing and verification work, and for something like a 6-month endurance test on an avionics box this just plain takes a lot of time and money. I'm not sure that my friend was accurate, but it seems reasonable to me that that kind of thing could cause big delays.
  - humbledrone 15y ago
    Obviously, any original estimate for how much money and time it will take to create the aircraft should include a reasonable level of buffering to account for the "known unknowns." Companies like Boeing and Lockheed aren't exactly newcomers to the aviation game, and I'm sure that they have extremely detailed statistics on how common this setback is, how much it costs on average, etc. It's the same way in software -- when estimating how long a large project will take to complete, you have to assume that some things will go wrong. Maybe you have to track down a nasty bug in a third-party library, or you lose a team member, etc. All of this has to be considered based on its impact and likelihood as part of up-front estimation.
    - hef19898 15y ago
      Yes, the known unknows. Everybody knows about them, but strange enough we all seam to be ill-prepared for them. You just have to look up new aerospace developments of late: F-22, JSF, the Kiowa repalcement, the new tankers, A400M, NH90, Eurofighter, Rafale, B787, A380, B747-8, A350... you name it. Almost all of them were over budget, late and ran into serious technical and performance issues. Not all of them were due to design changes. And even if, when you know that certain changes are going to blow your development apart you just don't ask for them as a customer. A good counter example is the F-117, that one was, aparantly, in budget, on-time and did what it was supposed to do. My theory is that it's the grown bureaucracy that it messing it all up. In the F-117s development there was none since it was more than just top-secret, they took what was available and turned out good. I know that one example doesn't make a proof, but it's pretty cose for me.
Symmetry 15y ago
Um, you're wrong. Most military or government aerospace development is done on cost plus contracts where if the bidding entity ends up spending more than they officially planned they'll make more profit. This isn't the way your city buys a sidewalk, but it is the way the government buy a space launch.
- krschultz 15y ago
  Not exactly. Non-recurring engineering is often cost plus. You simply can't scope that out very well. Purchasing more units of an already developed item is generally fixed price. However, even with cost plus there are incentives to come in faster/better/cheaper. If you run a company and have the choice between '1 project scheduled for 18 months, stretched to 2 years and collect the extra' or '2 projects scheduled for 18 months, each finished in a year, and collect all the extra bonuses', you typically choose the second plan.

jonknee 15y ago

> There are no cost overruns in government contracts. Just changes in scope. You don't find out what they really need until you really get into talking with them. Umm, there are tons of cost overruns in government contracts (and in non-government contracts). Consider the James Webb Space Telescope for example. Or the F-35. Or about any software project. It's very hard to predict costs of new inventions.

angersock 15y ago
So, in regards to fighterjets and things at least... I've heard (and I'd love to hear whether or not this sounds reasonable) that one of the big problems is that you'll source a part like a resistor, ASIC, or something, and then for whatever reason it'll get changed--maybe a cheaper part became available and you were forced to use the cheaper part, maybe the part is no longer in production and has to be switched out, maybe the part just failed to function correctly. The effect of this, of course, is that you have to redo all of your integration testing and verification work, and for something like a 6-month endurance test on an avionics box this just plain takes a lot of time and money. I'm not sure that my friend was accurate, but it seems reasonable to me that that kind of thing could cause big delays.
- humbledrone 15y ago
  Obviously, any original estimate for how much money and time it will take to create the aircraft should include a reasonable level of buffering to account for the "known unknowns." Companies like Boeing and Lockheed aren't exactly newcomers to the aviation game, and I'm sure that they have extremely detailed statistics on how common this setback is, how much it costs on average, etc. It's the same way in software -- when estimating how long a large project will take to complete, you have to assume that some things will go wrong. Maybe you have to track down a nasty bug in a third-party library, or you lose a team member, etc. All of this has to be considered based on its impact and likelihood as part of up-front estimation.
  - hef19898 15y ago
    Yes, the known unknows. Everybody knows about them, but strange enough we all seam to be ill-prepared for them. You just have to look up new aerospace developments of late: F-22, JSF, the Kiowa repalcement, the new tankers, A400M, NH90, Eurofighter, Rafale, B787, A380, B747-8, A350... you name it. Almost all of them were over budget, late and ran into serious technical and performance issues. Not all of them were due to design changes. And even if, when you know that certain changes are going to blow your development apart you just don't ask for them as a customer. A good counter example is the F-117, that one was, aparantly, in budget, on-time and did what it was supposed to do. My theory is that it's the grown bureaucracy that it messing it all up. In the F-117s development there was none since it was more than just top-secret, they took what was available and turned out good. I know that one example doesn't make a proof, but it's pretty cose for me.

Symmetry 15y ago

Um, you're wrong. Most military or government aerospace development is done on cost plus contracts where if the bidding entity ends up spending more than they officially planned they'll make more profit. This isn't the way your city buys a sidewalk, but it is the way the government buy a space launch.

krschultz 15y ago
Not exactly. Non-recurring engineering is often cost plus. You simply can't scope that out very well. Purchasing more units of an already developed item is generally fixed price. However, even with cost plus there are incentives to come in faster/better/cheaper. If you run a company and have the choice between '1 project scheduled for 18 months, stretched to 2 years and collect the extra' or '2 projects scheduled for 18 months, each finished in a year, and collect all the extra bonuses', you typically choose the second plan.