I worked on a 9-figure construction project and there were two full-time engineers specifically tasked with looking for ways to reduce the cost of the project, because it was essentially free money for the contractor. These two people generated millions in profit for the company.
In the end markets simply aren’t efficient without pricing information. If you can’t get someone to bid a fixed price you might as well keep in in house.
As to failing to finish a project the wonders of fixed price contracting is you can pay at completion of some milestone. Of course that means people need to pad the price even more ahead of time to assume that risk, but price becomes a really useful signal.
Semi-related story:
Way back I was a trainee for a company, which won a contract to upgrade the the branch servers for some other company, about 500 of them.
The on-site process involved copying over files from the old Unix server to a new NT4 server running Citrix, split maildir and home directory and set proper user-level permissions on each users home and mail directory.
The whole rollout was going on for just a few months, so we had tons of folks doing this in parallel at each site. During the first site I saw the manual our project team had created, and it specified us doing the copying and permission setting using Windows Explorer. To change the ACL that meant scrolling down to find the right abbreviated user name matching the directory name... fun when there was hundred employees at a branch.
I figured there had to be a better way, and started writing some batch scripts. After about a day I had automated the whole process, including sending a mail to our phones (the phone company had a service turning [email protected] into an SMS).
So I just launched the script, went for a 1-2 hour lunch until the phone beeped. Then I just had to go over the logs and verify there were no errors, and job done.
Back at base, I shared the scripts with my coworkers, which obviously loved not having to do the tedious process as specified.
It was only after the end of the project that I learned that the contract included a fixed rate of 3-5 days per site for one or two techs, depending on size. After all, changing all those permissions manually took a lot of time. After I wrote the scripts nobody used more than half a day...
Did I get any extra compensation for that? Of course not, and I was too young and shy to ask.
Oh well, lesson learned.
Fresh out of high school, I worked for a telecom contractor, installing large cross-connect machines. (Think 20-30 racks of equipment, a few thousand DS3s worth of capacity, which in the late 1990s was a lot.) We'd do the physical install and some very basic continuity testing, then hand over the keys to the vendor's own tech who would "turn up" the machine, thoroughly testing it and making it ready for service, before finally handing it to the customer who would provision their actual traffic on it.
The turn-up procedure was documented in the manual and technically anyone could do it, it was just customary to have the vendor do it. And they charged handsomely for it, flying out a tech who would spend 2-3 weeks on site running through the procedure. It was quite a lot of manual work, adding and deleting connections throughout the whole range, and plugging a T-Berd tester into the patchbay to verify transmission of every single circuit. The tech was quite busy during those weeks.
My company decided (at a level way above me) that we were gonna try to get into this line of work. As the pimply-faced youth with some computer experience, I got tapped for it, and paired with a graybeard who had done something similar back in the 80s and had a sense of the lay of the land. The logic was that our first few jobs might not make us any money, but we'd get better, and anyway just saving the airfare would be a win.
The manual specified typing all the commands directly into the machine's (TL1) prompt, but I wanted to avoid typos, so I typed everything into Wordpad first and then pasted it into the prompt. Which meant I was creating a log of all the necessary commands, and I could search-and-replace various parts to advance through the range of valid circuits. Less typing, more pasting. By the end of our second day, we were screaming through the testing (still doing everything by the book, just typing an effective 500wpm and making zero typos) way faster than the vendor's tech ever did. Running back and forth with patch cords and T-Berds was the principal limit.
By the end of the first week, I had written a QBASIC program to obviate the search-and-replace, and just spit out bespoke commands for any given range, or for a whole machine at once if you wanted to feed it the whole config. The output was a many-thousand-line text file which you would then drip-feed to the machine while doing the needful with patch cords and your trusty T-berd. As long as nothing went wrong, it was literally just one long file-send with pauses for jogging across the office. And as our installers were very good, things rarely went wrong. (I did find out that the machine's input buffer was only a few hundred characters, thank goodness the terminal software had an inter-line paste delay!)
Our very first job came in at less than half the time we'd bid, and it only improved from there. After our third job, I believe, I went solo. I bought a pair of WPC11 cards and parked one laptop at the machine's console port, carried the other with me to the patchbay, and eliminated the running-back-and-forth time too. Wizardry! Laziness! Profit!
I proceeded to reverse-engineer quite a lot of that machine, finding undocumented debug interfaces that saved tons of time when things did go wrong, so we could avoid the blind combinatorics specified in the manual, and instead directly zero in on the errant signal. When things went smoothly, I was 3-5x faster than the vendor's own tech. When things went wrong, I was 10x faster.
Extra money? Nah. But I was having the time of my life.
And/or some vendor "engineers" have a more "follow the book" and less initiative to make things better.
Obviously this example is not a real world one as construction is not my area of expertise but I do think I have insights into mundane human mendacity. Then again I am also prone to cynicism.
On a related note to using the wrong concrete, things like this can be an opportunity to see master craftspeople in their element. We had a truck of concrete show up and the foreman rejected it with almost no hesitation. "Why the f*&($ won't you sign on off this!?" the driver yelled. "It's smells wrong", said the foreman. Nobody else noticed anything amiss, but he was the boss, so the truck went back to the plant and sure enough, they'd mixed it wrong.
I think the offset is to have a mixture of QA/checkers and gov't regulators who are onsite checking things. With huge civil engineering projects, sometimes the gov't will hire a competing civil engineering firm just to be onsite checking the work. That balance probably works.
a) Unless very carefully managed, you can end up with contractors being incentivised to cut corners (i.e. We make more money if we buy the cheapest doors possible, regardless of if they only last a few weeks).
b) If contractors have fixed profits at their bid-price, they will have zero incentive to accept any change requests (unless they come with additional margin, in which case, overspends will magically turn into change requests).
This isn't software where you have hand-wavy goals written by people with no training and can refactor when your tests fail :)
If the specification is that precice, the best fit is usually a fixed price contract.
Typically open-book contracts is to account for uncertainty / flexibility / changes without having to raise endless CRs/WOs. Those fixed price contracts can still have gainshare mechanisms to identify joint cost saving.
The worst example I can think of would probably be the nuclear industry where operators providing grid power on cost plus contracts teamed up with environmentalists after Three Mile Island to get laws passed insisting that nuclear power had to be as safe as possible regardless of cost, which turned into essentially a license to print money.
The entire point of getting the private sector is that their insentive should be to do it in the most efficient manner possible, and noone has to manage them - the inefficient ones just die out.
The second-best approach is to have the government manage the whole thing, where they have no real incentive to be efficient.
But what we have here is employing the private sector and incentivising them to be as inefficient as possible, and the government must still hire an army of accountants and experts to keep them in check. it's combines the worst aspects of both!
It seems like a common misconception that businesses do anything in 'the most efficient manner possible'. A business's incentive is to do things in 'the most profitable manner possible'. It happens that efficient and profitable have a lot of overlap, and business schools love case studies about eliminating inefficiencies, but in the real world it is often easier to boost profit thru marketing/client relations/lobbying than thru searching for possible inefficiencies.
Not all contracts are automatically won by the lowest-cost bidder. Yes, it allows corruption on the part of the person doing the hiring, but it's better than the alternative.
Which of many possible alternatives and how do you demonstrate that it's better on any of the areas where tradeoffs are made?
Having a government managed but privately delivered program, having a totally privatised program, having a completely government run program?
Having the private firm take on the risk instead of the taxpayer? having the risk split 50/50?
What happens is that they get promoted to another team, or put on a project that has absolutely no future, and everyone knows it.
Why is it difficult to remove people? Because in the old days, newly elected executives would clear out the civil service and hire or contract with people that supported their election campaign.
I have wondered if we could ~endow government agencies in a way that would let them fail, and do the same to spin up ~internal competitors when we feel like turning up competition/innovation.
But it's not something I have gamed out enough steps to feel confident about...
> the private sector is that their insentive should be to do it in the most efficient manner possible
Becoming somewhat more common now is a hybrid approach of cost-plus/fixed. Contracts for development phases up to a certain milestone are set up as a cost plus contract and then production comes under a fixed price contract.
https://whistleblower.org/whistleblower-profiles/thinthread-...
Part of the problem here is that the government actors selecting the contracts also want more money flowing through their agency, and they seem to get kickbacks in the form of 'retirement positions' with the very private contractors they're directing contracts to.
Anyways, this was the explicit reasoning of the ACA drafters when they included the profit cap. Maybe they were wrong, but this was thought about & discussed a great deal at the time
Competetive bids drive costrs down. The alternative to cost plus is? Not having fixed profit margins, in which case there is still incentives to increase costs.
In both cases the limit to costing is competition and the contracting side.
Another way to think of it is cost plus lowers costs by lowering risk. If the company were on the hook for uncertainty (which is the case on ANY project), then they would have to price that into the original proposal, as would their competitors, ensuring more costs to cover the risk.
By lowering risk to companies, it provides another avenue to lower costs. This happens in lots of markets - look at cost margins where there is volatility versus margins where there is predictability.
They also get a fixed profit ratio relative to capex. So they burn giant piles of money to get their percentage from the government…
In other words, it is the “spherical cow” of contracting.
Why would this possibly be true? Two groups can bid on the same project with different tradeoffs. Projects are complex, and there are zillions of tradeoffs to make a competitive bid. The group offering the project then evaluates the bids, often with multiple rounds of feedback.
And generally then these projects initially got to several competitors, with fewer getting follow-on work, causing them all to try really hard to deliver good work.
Future projects are always worth more than current projects, so only an idiot does bad or over-cost work often. Such groups eventually don't get more work.
I've worked on such bids for decades. Multiple groups bid, with a statement of work. The other side evaluates the bids. Cost plus simply allows groups, when bidding to not have to use larger multipliers to mitigate risk, since some of the risk goes to the offeror.
If you were running a company, and you had two ways to bid a contract, firm fixed price vs cost plus, which do you think provides less risk to you?
And how does a company account for risk?
Higher prices, end of story. This is simple economics.
It's why less risk results in smaller margins throughout pretty much all industries. It allows stable planning.
Higher prices, end of story. This is simple economics."
The job of the company is to manage risk of the project they are delivering, and let the ones that do it well win. What the hell is the point of private sector if the taxpayer pics up the tab every time something goes wrong?
I don't know what bids you dealt with, the ones I did looked nothing klike what you describe. For any non-mundane work, that's not like, paving a street, you low-ball the bid, deliver the bare minimum the paperwork allows and then keep milking the government for improvements because what you've delivered is not good enough.
If the bid is about new technology, the tradeoffs cannot be understood by non-experts, and are difficult to pinpoint even for experts. The expertiese required to access how good is the bid is the same as the expertiese needed to design the system -> if you have that expertiese on hand, you would not need to outsource the design.
The process is insanely time intensive, expensive, and difficult to navigate. Not because government is inefficient, but because taht's the nature of new technology.
> Future projects are always worth more than current projects, so only an idiot does bad or over-cost work often. Such groups eventually don't get more work.
In UK Capita get work again and again despite fucking it up all the time
Contracts where the contracting party agrees to a % of responsibility. When the project runs over the profit is reduced by the overrun * the percent responsibility.
Lets say the agreement has 25 million in profit and 40% responsibility. If the project runs over by $2m, the contractor eats $2m * .4 = $800k reduction in profit so now it's $24.2m.
Then they're actually incentivized to not have overruns.
Except no big agency does this because it's a legal nightmare. The others (cost plus and firm fixed) are well explored.
>Then they're actually incentivized to not have overruns.
Everyone has incentives not to overrun - it means you're less likely to get future work, and future work is worth much more than overrunning current work.
agencies are starting to do it more and more often. It won't get explored if people are scared, but it will get explored if people actually want to fix the problems with procurement overruns.
>Everyone has incentives not to overrun
in cost plus your overrun often directly leads to you making more money. So no, not exactly.
>it means you're less likely to get future work, and future work is worth much more than overrunning current work.
Maybe, maybe not. Lots of things have few competitive bidders, or bidders who bid knowing they're going to make overruns and claims to make the contract profitable when their bid never was. it's often called 'buying the job' in other industries.
Depending on program goals, both approaches have their place. However for projects like Artemis which tried to minimize R&D cost and save time through component re-use, the latter approach would clearly be better.
Are they though? If a casual layman internet reader can see the problem and see that they always get taken advantage of for profit, couldn't also the NASA officials and contractors signing these contracts? They know exactly what they are doing.
SpaceX signed non-cost plus contracts and delivered far more quickly and cheaply than the rest.
To further this, there is also no incentive and funds to modernize an already existing product. If it was written in Java 6 using J2EE, it’s going to continue to be that until it dies.
Folks I worked with also didn’t care to up skill or learn new ways of doing things because the incentive wasn’t there. Their cost was fixed, their price was fixed, their profit was fixed. They will do the bare minimum to satisfy the contract.
However, people aren't usually a fan of them because it requires all of the overhead/audits of a cost-plus contract which typically minimizes how cost effective they can actually be at getting below the target price.
There are also some development-type projects where the government will fund a project, but still require some amount of funds (like 25%) to come from the contractor. But this is really early development type stuff. Basically it is a way to influence where a company is spending their R&D money. Its not really for any operational systems.
The only good answer I've heard is for the agency to have enough technical experts that they can spot bad things. However that means the government needs to develop technical experts.
It's probably beneficial for the contracted company to have technically inclined individuals in leadership too, ranging all the way from the lowest levels up through CEO. Engineering-minded management is much more likely to spot potential troublemakers and risks for runaway costs in the planning phases where someone with a background in management or finance is more likely to fudge things and say "yes we can do this at X unrealistic low cost" just to win a bid. They're also less likely to make bad calls on how to fix budget problems that crop up midway through the contract.
Make companies evaluate and eat the risk by allowing them to set the fixed price...
> yet incentivizes minimizing that cost.
... where they are bidding on the contract against other companies.
This only works if you have a market with multiple players, which is possibly the main reason NASA is more heavily leaning towards fixed-price contracts now.
its better to have trusted suppliers and relationships that lead to long term waste (cost) reductions and quality output (you know your client and your client knows you)
deming said it better than i can: https://deming.org/minimize-total-cost/
But first company will then go out of business or become efficient.
Survival of the fittest evolution style for business.
The new company may be cheaper or have better quality but that isn't an automatic win. The second company needs more money or an event to superseed the first story.
For some definition of well intentioned!
But in practice, this means that in order to increase profit companies have to increase costs. If you're building a product and you can spend $10 million to build it efficiently or $20 million to build it inefficiently, it's actually beneficial to do the latter. It's disincentivizing cost reductions and incentivizing cost increases.