Jump to content
The Wifcon Forums and Blogs

Witty_Username

Members
  • Posts

    33
  • Joined

  • Last visited

Recent Profile Visitors

370 profile views
  • cds

  1. If it's designed correctly to take over the dull, dirty, and dangerous tasks (ok, maybe 1102s only deal with the dull, not the other two) and then to feed the right information to the 1102 at each decision point for their decision it could be great. AIs have a relative advantage in some areas and humans have a relative advantage in others (and those areas are likely to change as AIs get more advanced). But imagining giving the AI a big database of existing government contracts to learn from reminds me of this: Saturday Morning Breakfast Cereal - Rise of the Machines (smbc-comics.com)
  2. Interesting. In that case you might more broadly define the need as finding out whether some particular design will work. "Failing fast" could be seen as fulfilling the need to find out what works, make changes, and try again. Of course you can keep opening the aperture and question whether any particular platform satisfies the need for a specific capability, and whether any specific capability satisfies the need for national defense, etc. I certainly think there is likely to be value in speeding up PALT, however it is defined, as long as it isn't done at the expense of some other part of the overall process. Understanding where contracting (or any particular role) fits in to the overall process is a challenge, particularly in a contracting office that supports many different external "customers", but I think by trying to understand the goals of the level above you (e.g. contracting understands program goals, etc.) you're more likely to be able to optimize your own part of the process in a way that doesn't negatively affect the overall process. Alternately if you don't understand the higher level goals you may well optimize your own part of the process in a way that hurts the overall process.
  3. I agree. And maybe the PALT on some of the DIVADS or A-12 contracts was 72% faster than current PALT, and yet the need was never satisfied. So I wouldn't declare overall success based on how fast a contract was awarded that never actually fulfilled the need. But, easy for me to say, hard to actually do (except to be aware that we shouldn't overly-incentivize a narrow definition of PALT that doesn't include delivery/success)...
  4. Solicitation to award is a very industry-focused definition of PALT, because it represents the time when offerors have their resources tied up in a proposal. It's probably not particularly meaningful to a program office or contracting office at all. Incentivizes contracting offices to minimize proposal preparation time and to award without discussions. The time from receipt of a "procurement-ready package" to award is a contracting office focused-definition of PALT. Also probably not very meaningful to a program office who can't get a procurement-ready package in the door. Incentivizes contracting offices to not "accept" less than perfect packages, which may well actually increase overall requirement satisfaction lead time for the program office. The time from identification of a need through delivery is a program office-focused definition of PALT and probably what really should be incentivized. But it's too broad to really assign responsibility for to a single entity (assuming the program office and contracting office are separate) so it's probably hard to do.
  5. I don't think it's helpful to try to categorize every requirement as either a service or a supply. I think the must useful categorization of a license or subscription, or an equipment rental or short-term facility rental (all of which have alpha character "service" PSCs but don't really fall within the definition of a service) is "not a FAR Part 37 Service." I'm not saying it is a supply, I'm just saying it isn't a service in the FAR Part 37 sense of the word where an identifiable task is performed by a contractor employee (or the T. Peter Hill, "On Goods and Services", 23 REV. INCOME & WEALTH 315, Dec. 1977 sense of the word as a "change in the condition of a person, or of a good belonging to some economic unit, which is brought about as the result of the activity of some other economic unit, with the prior agreement of the former person or economic unit”; or Vern Edwards, "Performance Work Statements", The Nash & Cibinic Report Vol 35 Issue 2, Feb. 2021 sense of the word as "work done in order to change an existing state of affairs" either). Why is it useful to separate from FAR part 37? Because there are a bunch of extra things you have to do for a FAR part 37 service that don't make sense for a non-FAR part 37 service, like a determination of non-personal services, inherently-governmental determination, acquisition strategy for the acquisition of services, performance-based service acquisition, maybe pay in arrears, etc. and yet it still doesn't make sense to call an intangible license a "supply." The Section 809 Panel Recommendation 43 notes that it is "[unrealistic to categorize] all purchases as either supplies or services [...] often leading to contracts that are neither optimized nor appropriate for the solution being acquired." So, call it a non-FAR 37 service, use the alpha-character PSC, and create a contract optimized for the actual requirement, not any particular category.
  6. I think that's reasonable. If you look at the software PSCs the manual is clear that the difference between a D "service" PSC and a 7 "supply" PSC is whether the license is perpetual or not, and the same logic could apply in your case. I don't think the use of an alpha-character PSC as prescribed in the PSC manual makes it a service in the FAR 37 sense, but you may still run into issues with CAR coding (e.g. if it's not a service at all do you call it a is it a performance based service or not a performance based service?), and people who expect anything with an alpha PSC to be treated entirely like a FAR 37 service (payment in arrears, certification of non-personal services, QASP, etc.). But to paraphrase Vern Edwards quoting GAO paraphrasing Galileo: it doesn't matter what you call something, it matters what it is...
  7. I think any system that involves so many people (a quick Google search said 165,000 people in the DOD acquisition workforce alone in 2017) is going to display emergent behavior. Hire a huge workforce, give them a complicated set of rules and systems, then change the rules and systems continuously over their careers and I believe as new rules are individually interpreted, new concepts are adopted, old habits gradually fade, and new habits are formed you're likely to get very unpredictable outcomes.
  8. I read Kill Chain and found it relevant and interesting (I originally was going to say "I enjoyed it" but that's not quite right). I don't recall coming away with the impression that the DOD was necessarily against adopting AI (although within narrower areas it seems like pilots tend to think that aircraft need pilots, and tankers tend to think that the military needs tanks), but rather that congress, the military, and industry (all "in risk-averse bureaucracies that seem inclined to resist and stymie change at every turn") are too wedded to the idea of improving "platforms" instead of making more effective kill chains. I do think that there are plenty of people and organizations outside the DOD who are very concerned about including AI in the kill chain, and explaining why AI could make war more ethical, not less ethical, seemed to be a major focus of the book. The book has a great explanation and critique of the defense acquisition system; here's a key explanation for the situation for each major player. DOD: "The issue is not a lack of authority to go faster or take more risk, but that those who must exercise those authorities, bear those risks, and be accountable for the outcomes rarely use the authorities they have. [...] Those who do make decisions often lack the authority and incentives to make riskier decisions to get better outcomes." Congress: "Members of congress cannot ignore the things and issues that are likely to win or lose them votes. [...] The future does not vote, but the present certainly does. [...] Congress passed law after law to create new processes, offices, paperwork requirements, and official homework checkers to ensure that some bad thing that happened once would never happen again." (maybe he was paraphrasing from Vern Edwards “Frictionless Acquisition” Nash & Cibinic Report Sep 2020 Volume 34 Issue 9 “The principal acquisition statutes […] reflect congressional responses to various issues that have annoyed politicians from time to time.”) Industry: "[When there were no longer any existential threats and congress and the DOD began to focus on transparency, efficiency etc.] "The US government created incentives for defense contractors to do the wrong things, and that is often what happened. [...] Companies adapted. [...] Many companies resented making these changes, which they felt forced into. [...] Many of these major moves represented gains in efficiency, but often at a cost to the effectiveness and speed of innovation." So I felt like there were two major takeaways from the book: 1) We need kill chains with cutting edge AI instead of platforms if we want our military to remain relevant; 2) "It ultimately comes down to incentives" which are currently: short-term focus on spending money in their districts and avoiding the next procurement scandal from congress incentivized by voters; risk-averse compliance-focused behavior from acquisition professionals incentivized both by congress and internal leadership; and an industrial base that gives both parties what they ask for (replacement platforms and auditable accounting systems).
  9. I would say certifications (or education, or experience, etc.) are often used when there is no measurable objective for the contractor to achieve, as is often the case in professional support services contracts. So the logic is likely "we want someone to sit here and work on this, but we can't really define the outcome of the work, so we at least want someone trained/experienced/certified." I'm not saying it's necessarily a good idea, or a good way to contract, but that's generally the thought process I hear.
  10. I think it's worth considering that offerors make choices in their proposals, and if the government can tell them what they value the most it can help them make the choices needed to submit offers that the government wants. For example, if I want to buy a house I don't just tell the realtor "show me every house with at least three bedrooms" and then try to sort through the pile, I might instead say "A low price is most important to me, show me low-priced three bedroom houses" or alternatively "I want as much room as possible and I'm willing to pay for it" and now the houses the realtor shows me are more likely to be what I'm looking for, and I should have some similar houses to compare. When it comes down to choosing which house to buy I'll still decide whether the extra money is worth the extra space in the end, but I likely won't be deciding between a tiny foreclosure and a million-dollar mansion. Same thing when you buy a car, you don't go to a Ford dealer and drive a Focus, a Mustang, and an F-150; you tell them you are interested in a economy, or speed, or towing capacity, and then you drive some cars that meet your preferred criteria. You still make tradeoffs (is the additional towing capacity of an F-250 worth the price compared to an F-150) but they are showing you the vehicles that they think are most in line with what you are looking for. In the same way, if a solicitation says past performance is the most important factor and non-price factors are significantly more important than price potential offerors may decide to invest in a teaming arrangement with someone currently performing similar work, even if it is more expensive, rather than leaning their process to try to propose the lowest price they can. Now, the government will (ideally) get several proposals from offerors with solid past performance to make tradeoffs among, rather than a random collection of high and low priced offerors with widely varying past performance.
  11. If this is a supply contract I would suggest that if you know you will not drink the water until next fiscal year you don't have a bona fide need for the water this fiscal year. Supplies are generally a bona fide need in the period in which they are used. There are two exceptions to the bona fide needs rule for supplies: stock-level exception, which would absolutely apply if you purchase a year's worth of bottles and kept them in your fridge and drank one per month; and lead-time exception for deliver lead-time or production lead-time, neither of which seem applicable here. I'm not necessarily convinced that it is a supply contract though. Maybe you are really paying a service contractor to maintain the water supply in your various buildings as a service, not simply ordering a supply delivered, in which case if you have a statutory exception to the bona fide needs rule for severable services (10 U.S.C. § 2410a for DOD or 41 U.S.C. § 3902 for non-DOD).
  12. Looks like it is the same as "Agency Code" here: https://www.fpds.gov/downloads/top_requests/FPDSNG_Contracting_Offices.xls Here are some examples: “5700” (Air Force), “2100” (Army), “1700” (Navy).
  13. Not to hijack the thread, but on a semi-related note apparently DPC is saying as of March that to order from an existing Government contract a cardholder needs to be designated "by name" by the contracting officer who awarded the contract (source: https://www.acq.osd.mil/dpap/pdi/pc/docs/SmartPay3/Ref B - JAM GPC Role Descriptions_3-23-21.docx).
  14. This seems so basic, but I can't make sense of it: What does FAR 13.202 mean when it says "The clause at 52.232-39, Unenforceability of Unauthorized Obligations, automatically applies to any micro-purchase, including those made with the Governmentwide purchase card"? The interim rule in Federal Register Vol. 78, No. 120, subsequently adopted without change in the final rule, discusses the problem they are trying to solve (open-ended indemnification) and the proposed fix is "FAR 13.202, Unenforceability of unauthorized obligations in micropurchases, is added to require the clause at 52.232–39, Unenforceability of Unauthorized Obligations, to automatically apply to any micropurchase, to prevent violations of the Anti-Deficiency Act." That language, "automatically apply", is different from any other prescription I could find in the FAR, implying that it is not simply requiring that the clause be included in some type of order, but if it were not included in an order (or ordering agreement) through what mechanism would that clause apply to a vendor? Maybe it was just strangely worded (although the use of "automatic" wasn't a one-time thing, it shows up multiple times across the interim and final rules before ending up in the FAR) because later in the 2020 final rule implementing the increased micro-purchase threshold it says "The additional awards at or below the MPT would not require provisions or clauses, except as provided in FAR 13.202" which seems to acknowledge that an award below the micropurchase threshold would, in fact, require the clause as provided in FAR 13.202, and so maybe it is not "automatically" applicable at all?
  15. And to throw in a Tragedy of the Commons reference: while any increased risk of a protest accrues directly to the contracting officer, any corresponding future benefit of generally improved proposals by the company is spread across all contracting officers. As an aside I think this separate allocation of risk and reward is reducing the usefulness of past performance evaluations too, because each individual contracting officer is disincentivized to actually report poor or even average performance since they risk a challenge that increases their individual workload now, while the corresponding benefit of having accurate past performance information available for future source selections is spread across all potential contracting agencies. But back to the specific question of whether more or less information actually increases or decreases protest risk I think Vern Edwards said it best in a post on discussions (http://www.wifcon.com/arc/forum355.htm): "If you don't do first rate evaluations, then maybe you should hide the evidence, at least until the GAO or the Court of Federal Claims asks for it and gives it to the protester's lawyer. If you have incompetents conducting your business, then maybe you should hide their work as long as possible."
×
×
  • Create New...