Nice words; let's see the impacts before we judge. Feels like we've heard transformation a million times. I also find it rich that the government forced - forced - the defense industrial base to consolidate and now condemns the defense industrial base for being consolidated and is going to punish them with new rules (procurement rules, stock buy-back rules, etc.) the others don't have to abide by. Feels like empty calories - delicious but not nutritious.
In fairness, defense industry consolidation was an inexorable consequence of the end of the Cold War and the drawdown of all military capabilities during our decade-long "holiday from history."
Both GWHB (aka "Bush 41") and Clinton contributed mightily to this. "Peace Dividend"...ptui. As one of my Air Force colleagues put it, "The 'Peace Dividend'...is peace."
But even before the Cold War was well and truly over, contractors were getting out of the business. I worked at NAVAIR in air-to-air and air-to-ground missiles during a period when all the expensive, painful work we'd done dual-sourcing almost all our missile systems was erased by a massive industry implosion.
TI sold their missile division (HARM and Paveway) to Raytheon. GD sold Convair (cruise missile) and their tactical missile division (Sparrow and [IIRC] Standard) to Raytheon as well, as did Hughes (Phoenix, AMRAAM and AAAM, though the last was still in development). Ford Aerospace sold their Sidewinder division to...can't even remember the name of the buyer anymore.
So in a few short years all that industrial base went from a half-dozen to basically one.
That's just market forces at work. When we stopped buying Phoenix--I awarded the very last production contract in 1993--there was no reason for Hughes to stay in the business, and Raytheon wound up moving most of their production lines from Massachusetts to Arizona...
You see the same forces at work in both airframes and shipbuilding. But--again--in both those cases there were market forces at work. No vast conspiracy necessary! :-)
I make no claim that there weren't things going on behind the scenes. But the only "force" that was at work when I was there (1981-1996) was the market--which consisted of a single monopsonistic buyer.
I don't disagree with you - but it was the government who purposely ordered these guys to all consolidate. It wasn't the "invisible hand" of pure market forces. To your point - these are the dividends we have reaped from the so-called peace dividend. So here we are.
Well, I tried to make it clear that I wasn't talking about a classic "invisible-hand" market. But a market is a market even if there's only one buyer.
Indeed that's why we went to all that effort to establish--in the phraseology of the day--"balanced and *geographically dispersed* sources of supply." [my emphasis]
Hughes/Raytheon; Ford/Raytheon; McDonnell/Convair; the list goes on. And of course you saw that in shipbuilding as well: EB/Newport News; Litton Ingalls/BIW. Probably others I can't remember.
But I wonder whether the government *forced* the consolidation or merely *facilitated* it...by telling the players that they didn't have to worry about the FTC or anybody like that getting in their way.
...Can you provide a pointer to your source for the "forced" thing?
"With the end of the Cold War, the U.S. government assessed that its military requirements had diminished and began to reduce defense spending. Because many firms in the commercial DIB were heavily reliant on the defense market, DOD determined that the sector's continued viability depended on restructuring. Accordingly, the government actively encouraged companies to pursue consolidation, with the result that by the early 2000s the number of prime contractors had diminished from 51 to 5."
I will concede that "forced" could be taken as too strong - but when your sole customer "actively encourages" you to consolidate, what choice do you really have?
Hi Dave. I was the guy at NAVCOMPT that got SECNAV Lehman to award a second source for Phoenix to Raytheon to get annual head-to-head competitive savings like on all my other WPN procurements. Five-year contract that I then canceled after three-years because Navy changed its Phoenix employment tactics to NOT shoot over the horizon at Bear bombers. Had to pay-off the unreimbursed non-recurring special tooling and test equipment, but three-years of savings from both contractors paid off and reset baseline for original prime.
I did an around-the-country visit to all missile manufacturers. Ford was past their prime and their facilities (in Newport Beach, CA) and managers showed it. I was impressed with Raytheon guys in MA and VP told me his best engineers liked to work test and evaluation shop testing and stressing their systems to try and break them, using feedback to improve their missile. The Tomahawk GD guys in San Diego with every-station feeding their status, progress and problems into master control database in real time was impressive. Tomahawk is my "Harvard business school case study" in head-to-head competition (what I call "super competition"), where annual competitions result in better overall results than a single, winner-take-all five-year multiyear. 30 percent unit cost savings from NAVSEA cost-estimator baseline by year five! To get program office to agree, I made a deal to give them back half the savings to buy more Tomahawks (going from 360 a year to 400) and giving the other half savings to SECNAV for other priorities.
I never worked with Ford Aerospace--I worked at various times in my career with TI, Hughes, Raytheon and GD Pomona--but yeah, your perceptions of Raytheon certainly accord with mine.
I never worked on Tomahawk but after the direct-report Tomahawk PM got put under NAVAIR's aegis, I heard a lot about Convair from my friends who had moved over.
Hughes were a weird bunch since they were effectively a not-for-profit organization (being a part of Summa Corp). So they were pretty high on the technological side as well. The thing that made them strange was, they had all these different manufacturing divisions, like little fiefdoms, and they hardly subcontracted anything...except to themselves.
Possibly the most interesting contractor was Grumman, who I worked with in my first trainee rotation. They effectively considered themselves an extension of the Navy, and this was exacerbated by--as one of them put it to me once--"Grumman don't care about profit, we jus' wanna keep da lights inna plant burnin'!" (Spelled as heard in that harsh Long Island accent).
If you ever get to St. Louis, visit the Prologue Room in the Boeing building at Lambert Field. It is open to the public. They have this entire wall that is probably 60 feet long with the Boeing history of aviation. It is a physical manifestation of defense megacorporation consolidation, showing all the companies they have gobbled up.
CDR Sal, thanks for the post, I missed the speech. Your words: " in the end there are significant, long-dwell systemic issues with raw material, industrial capacity, and workforce that will slow things down unless they are addressed. Iron, coke, chromium, and steel—and the people that put them on rolling stock to shipyards and factories—are suffering under the second-order effects of post-industrialization and low demand." Truer words seldom spoken. I'm gun shy of "transformation". Worked "joint experimentation" for a number of years, it failed, and I've got the receipts. Was victimized to put together the history of ACOM / US JFCOM in 2011. Six months of my life I'll never get back. If we approach "transformation" the way JFCOM did, it's time to start learning to speak Mandarin.
We are good at telling ourselves what needs to be done. However, Clausewitz reminds us that commanders have a finite amount of will. "He recognized that real-world wars are limited by politics and resources, meaning "will" (resolve, desire, spirit) is finite and varies, leading to a mutual escalation until the cost of effort outweighs the political goal." I'm told that President Reagan kept an index card in his pocket with the three things that mattered (to which he applied his will). Someone else would have to worry about everything else. So, whose index card is this written on?
It would be helpful to know who wrote the speech, whether a speechwriter or an actual policy person with some imagination to suggest how to see change through. Hegseth does not possess the intellectual acuity to draft a serious document. The suggestion that a passed over National Guard major has morphed into a 21st century Clausewitz of strategic strength beggars the imagination.
Wait…what? Someone needs to brief me on what was created that required a pushback against a “woke” and “social justice” AI. There is a backstory here I need to see.
A case of GIGO from what I can surmise. There are several famous cases of AI requests that returned unexpected results. Such as a request for "A picture of a WWII German soldier" getting a fairly accurately uniformed and armed soldier who was clearly of black African lineage. In essence the AI algorithms had been brought up woke so leaned to non-white like a TV commercial.
That was among the more egregiously stupid ones, but apparently text responses often ignored clear historical data and returned a woke slant to many things. Essentially, the database was corrupted with sources that had a slant to them.
Wish I had good sources to cite, but I tend to studiously avoid AI. Sort of feels like consorting with the devil to me. I'm sure a search will turn up some more on the subject.
AI did not even understood that there are some "historical data". AI is a statistical-based, pattern-seeking system. Its logic is not linear, cause-and-effect one, that humans have. Its logic is based on generalized statistical patters.
For example: for humans, 2+2=4 is logical, because if you took two apples and add two more (cause) the effect would be four apples. For AI, it would merely be "most likely 4", because statistically 4 is the answer most likely to be correct.
That's the reason for black German soldier in AI generation. AI have generalized soldier, and generalized German uniform. Generatling the image, the AI essentially generate a big field of random pixels, and seek in this randomness statistical similarities that correspond with the prompt ("A picture of a WWII German soldier"). Problem is, AI did not actually comprehend the casuistic interconnections here; it have no reason to seek specifically white Caucasian (yes, this term is outdated... but the whole racial theory is currently a mess).
Well, I get the lack of understanding of history and application of logic in AI*, but I think that this goes further than that. Certainly we can agree that historically the majority of soldiers were not black. That applies certainly to Europe and Asia, naturally much less so to Africa. In North and South America it is less applicable as well. Not sure about the demographics of South America, but certainly in North America the portion of soldiers that were/are black is significantly less than white though maybe equal with some other minorities.
The point of that being somewhere in the database the system is taught that soldiers are typically black when they are not. I suspect this is in large part due to the anti-war propaganda from the 60s on that says something to the effect of "black people are forced to fight and die in disproportionate numbers here white people avoid combat and the military all together (with an added caveat in the last decade "because of white privilege.)"
So still GIGO. The data given is selective. It would be interesting to see if an AI could actually manage a database of something close to all the world's knowledge.
Which raises yet another interesting point: Can AI database compilers add copywritten material without express permission? And what are the ramifications of having, or not having, such things in the database? While copyright covers a lot of ground something as simple as a book on a historical subject might be denied to the database when the contribution to gathering correct answers could be significant.
*Though if taught logic, I think that could be done as well...though applying if/then, and/or/nor logic to machine learning is likely very difficult in any but micro cases as the means of resolving takes some value judgements which is but the flip of a byte in a computer.
The thing about AI is, it's the right tool for a job, somewhere, but not for every job. Particularly, the following is not a use case for AI: getting factual answers to simple questions. AI doesn't do facts. It takes this or that piece out of its training data. If the training data is crap, the answer will be crap. If the training data is excellent, AI may still screw it up. The kind of AI we're all using now (large language models) has no concept of truth vs. falsehood, and cannot have that concept. Can AI see trends in data? Sure. Can it select a reasonable course of action from a constrained set of possible actions? Probably, but that should be as a decision aid to a human, not in place of the human. Part of AI research needs to be figuring out something useful we can do with it.
It seems like one of the most important things about how one uses an AI answer is understanding the database that it draws from.
However, due to the vast amounts of information that may be in such databases it seems likely that it would be hard for any individual user to truly know and understand the knowledge base.
One would have to take the designer's word/notes on it for what was included and the opinions that it would follow. I personally would be leery of accepting assurances from anyone I was not familiar with unless I did some checking myself. Cf.
Well, considering how the first couple years of the $1.5 trillion defense budget will have to go to infrastructure (nitrocelluose/gun cotton, artillery shell/missile production lines, shipbuilding/repair facilities), plenty of steel to be brought from the US metals industry.
"We killed the old defense acquisition system and created the warfighting acquisition system to focus on speed, risk and accountability. With the AI strategy and innovation ecosystem transformation that we have just outlined, we are welding that third piece into place."
....this 'old' guy is just under 18 Mondays from departing the 'bidness' for good BUT, while I'm still lending a hand, I haven't seen anything at the program office level indicating JCIDS is dead.....since my entry into this mess in 1983, "transformation" and the search for "speed to the Fleet" and other associated phrases haven't yielded what I'd call increased speed of product delivery.....I'm sure my cynicism is shining thru in this comment but let's not forget Newton's First Law (Inertia) - it is real and compounds with the 'this too shall pass' mentality (those acolytes raised in and nurtured by the system tend to dig their heels in)......in some ways, I'm kind of sad to be retiring as the coming three years - maybe 7 if the republicans can hang on - might be interesting within the five sided wind tunnel and the SYSCOMS.......
"There is another gem that demands a backstory, “data hoarding?”" Oh come on Sal, I'm sure you've seen it, Gods know I have. "Oh the Airforce did a study about that awhile back, but we can't seem to get the results."
"Yeah John's the only guy that knows anything about that system, you have to talk to him."
Between the guys that decide "knowledge is power, if I share the knowledge I lose power" and the guys that use knowledge for gatekeeping purposes, there's a lot of info hidden in desks and minds. Places where it helps an individual, but not the community. And it exists from the wrench turner on the waterfront, to the SES in his lofty office, 'far above the dirt and sound of work.'
I left honest service in the operating AF 41 years ago and began a sentence in acquisition. t that time "Packard" commission was doing the initiatives for reforming.....
I wonder if Grok will thunder through a program's "earned value" reports and recommend a system be terminated when its estimate at complete is 130% with less than 30% value earned.....?
That is if Grok gets data that is not lies.
In my time I saw more programs that needed cancelled than US could afford. They were not canceled.
First, cancel dogs, then talk about why the system failed and recommend fixes.
Often good engineering technique was ignored and tossed aside because it was not required by law.
Your $400 hammer is built to DoD specs.(prob. heavier than the hammer itself), bought by the Service AFTER the production run was finished and they bought ONE. In addition the Service in question bought NO spares or low-balled the buy during production(based on a flawed model of failure and or usage of said hammer). Yes, there is overhead, but cost is trapped in the Pentagon glacier known as the "procurement system". Reform is needed, program managers need to be terminated and Services can babble about STEM, but its time to put LOGIC(istics) back into procurement.
As I pointed out in yesterday's comment, SECNAVs over the pass quarter century have not suffered from a lack of gold sleeved flag rank or STEM accreditation, yet here is where all of that "expereince" and insider knowledge has led us.
The UK started forcing mergers of aircraft companies about 75 years ago. The result was a few bloated giants that were too slow and cumbersome to respond to an overly bureaucratic and redundant procurement system. Twenty years later with the notable exception of the BAE Hawk they could not produce an aircraft without a foreign partner.
We started down that road with the last supper. There's might be too much inertia overcome this.
On the other hand, French post-war aircraft industry merged just fine. Granted, the merge actually started pre-war with nationalization of aviation industry.
The French pre-war nationalization was an abject disaster. Post war, I believe many companies were privatized and then later merged. However, Bloch (later Dassault) still remains private.
Thanks for the great and concise synopsis, Sal, and good to read you're cautiously optimistic about it in its totality. (One caution, though: please hyperlink to a different source than Wikipedia for background info or bios on people. Wikipedia is complete trash). Keep up the great work!
"An optimist believes we live in the best of all possible worlds,...and the pessimist believes that is true. (someone's quote)" Regarding "transformation" and AI, we've seen this movie before, and we've gained no superior real advantage. After advising a recent commission charged with "reforming" the Pentagon's end-to-end resourcing/budgeting systems ("acquisition system") on problems and potential solutions, their report failed to address those areas. I wrote my own small super critical report stating that none of their 28 recommendations would result in any substantial improvements (in an administrative process supporting operational requirements) and sent it to the four congressional defense committees with 14 alternate recommendations. But lead out with my own experience: "I like to kid I only learned two things in grad school. Lesson One: Hire your superiors. (Which means, even if you know how to do a job, when you need something done, go find the person best qualified and experienced and hire them.) Lesson Two: When all else fails, reorganize (...because it has the appearance of progress)."
Reading the CDR's comments this morning I'm still a bit if a pessimist regarding all this chatter about reform and AI. Especially the part about AI sorting out all this unshared data. The entire contracting system doesn't improve because WE don't really want it to change. Yes, some improvement, but we can learn faster all the dumb stuff. We really don't need AI to tell us we need to field more munitions (even on expendable commercial type ships) to take out all the navy ships China is building? Or that we need to produce drones at scale (the Ukrainian army is already teaching that lesson to the Pentagon in real time)?
One of my alternate recommendations: Let industry self-finance ten-year contracts and produce efficiently at scale, with DoD buying it out at delivery. Keep DoD/Navy out of the shipyards. Start with MSC. Recapitalize the entire supporting fleet now. No more small contracts for one or two boats. (If anyone wants to see my alternate recommendations on really improving the PPBE process, send me an email at johnoking82@gmail.com.)
Nice words; let's see the impacts before we judge. Feels like we've heard transformation a million times. I also find it rich that the government forced - forced - the defense industrial base to consolidate and now condemns the defense industrial base for being consolidated and is going to punish them with new rules (procurement rules, stock buy-back rules, etc.) the others don't have to abide by. Feels like empty calories - delicious but not nutritious.
In fairness, defense industry consolidation was an inexorable consequence of the end of the Cold War and the drawdown of all military capabilities during our decade-long "holiday from history."
Both GWHB (aka "Bush 41") and Clinton contributed mightily to this. "Peace Dividend"...ptui. As one of my Air Force colleagues put it, "The 'Peace Dividend'...is peace."
But even before the Cold War was well and truly over, contractors were getting out of the business. I worked at NAVAIR in air-to-air and air-to-ground missiles during a period when all the expensive, painful work we'd done dual-sourcing almost all our missile systems was erased by a massive industry implosion.
TI sold their missile division (HARM and Paveway) to Raytheon. GD sold Convair (cruise missile) and their tactical missile division (Sparrow and [IIRC] Standard) to Raytheon as well, as did Hughes (Phoenix, AMRAAM and AAAM, though the last was still in development). Ford Aerospace sold their Sidewinder division to...can't even remember the name of the buyer anymore.
So in a few short years all that industrial base went from a half-dozen to basically one.
That's just market forces at work. When we stopped buying Phoenix--I awarded the very last production contract in 1993--there was no reason for Hughes to stay in the business, and Raytheon wound up moving most of their production lines from Massachusetts to Arizona...
You see the same forces at work in both airframes and shipbuilding. But--again--in both those cases there were market forces at work. No vast conspiracy necessary! :-)
I make no claim that there weren't things going on behind the scenes. But the only "force" that was at work when I was there (1981-1996) was the market--which consisted of a single monopsonistic buyer.
I don't disagree with you - but it was the government who purposely ordered these guys to all consolidate. It wasn't the "invisible hand" of pure market forces. To your point - these are the dividends we have reaped from the so-called peace dividend. So here we are.
Well, I tried to make it clear that I wasn't talking about a classic "invisible-hand" market. But a market is a market even if there's only one buyer.
Indeed that's why we went to all that effort to establish--in the phraseology of the day--"balanced and *geographically dispersed* sources of supply." [my emphasis]
Hughes/Raytheon; Ford/Raytheon; McDonnell/Convair; the list goes on. And of course you saw that in shipbuilding as well: EB/Newport News; Litton Ingalls/BIW. Probably others I can't remember.
But I wonder whether the government *forced* the consolidation or merely *facilitated* it...by telling the players that they didn't have to worry about the FTC or anybody like that getting in their way.
...Can you provide a pointer to your source for the "forced" thing?
"With the end of the Cold War, the U.S. government assessed that its military requirements had diminished and began to reduce defense spending. Because many firms in the commercial DIB were heavily reliant on the defense market, DOD determined that the sector's continued viability depended on restructuring. Accordingly, the government actively encouraged companies to pursue consolidation, with the result that by the early 2000s the number of prime contractors had diminished from 51 to 5."
https://www.congress.gov/crs-product/R47751
I will concede that "forced" could be taken as too strong - but when your sole customer "actively encourages" you to consolidate, what choice do you really have?
The alternative was to let them wither and die - without big government orders most of them wouldn't be able to survive anyway.
Thanks! I'll take a look! :-)
Hi Dave. I was the guy at NAVCOMPT that got SECNAV Lehman to award a second source for Phoenix to Raytheon to get annual head-to-head competitive savings like on all my other WPN procurements. Five-year contract that I then canceled after three-years because Navy changed its Phoenix employment tactics to NOT shoot over the horizon at Bear bombers. Had to pay-off the unreimbursed non-recurring special tooling and test equipment, but three-years of savings from both contractors paid off and reset baseline for original prime.
Cool beans! :-) I salute you, sir! :-)
...And...thanks! :-)
I did an around-the-country visit to all missile manufacturers. Ford was past their prime and their facilities (in Newport Beach, CA) and managers showed it. I was impressed with Raytheon guys in MA and VP told me his best engineers liked to work test and evaluation shop testing and stressing their systems to try and break them, using feedback to improve their missile. The Tomahawk GD guys in San Diego with every-station feeding their status, progress and problems into master control database in real time was impressive. Tomahawk is my "Harvard business school case study" in head-to-head competition (what I call "super competition"), where annual competitions result in better overall results than a single, winner-take-all five-year multiyear. 30 percent unit cost savings from NAVSEA cost-estimator baseline by year five! To get program office to agree, I made a deal to give them back half the savings to buy more Tomahawks (going from 360 a year to 400) and giving the other half savings to SECNAV for other priorities.
I never worked with Ford Aerospace--I worked at various times in my career with TI, Hughes, Raytheon and GD Pomona--but yeah, your perceptions of Raytheon certainly accord with mine.
I never worked on Tomahawk but after the direct-report Tomahawk PM got put under NAVAIR's aegis, I heard a lot about Convair from my friends who had moved over.
Hughes were a weird bunch since they were effectively a not-for-profit organization (being a part of Summa Corp). So they were pretty high on the technological side as well. The thing that made them strange was, they had all these different manufacturing divisions, like little fiefdoms, and they hardly subcontracted anything...except to themselves.
Possibly the most interesting contractor was Grumman, who I worked with in my first trainee rotation. They effectively considered themselves an extension of the Navy, and this was exacerbated by--as one of them put it to me once--"Grumman don't care about profit, we jus' wanna keep da lights inna plant burnin'!" (Spelled as heard in that harsh Long Island accent).
Good times. Good times...
I can hear that Long Island accent now. My father-in-law worked at Grumman on the F-14 program!
If you ever get to St. Louis, visit the Prologue Room in the Boeing building at Lambert Field. It is open to the public. They have this entire wall that is probably 60 feet long with the Boeing history of aviation. It is a physical manifestation of defense megacorporation consolidation, showing all the companies they have gobbled up.
https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhCAVYhhbyPnX6AcbN5dTAyZe2Jk_cD2IY5UVzGlPJ2kgMUerWPVXNpp7pvDKOZO3CK3hBEyJiGpyddApqnTYQJuajLRrIceswmp8w9fD59_pxdWjt3-MhXTYSe5DOKfkwyT7dnR-l1w04G/s1600/4a.JPG
Thanks! :-)
Thanks for this one. Important article
CDR Sal, thanks for the post, I missed the speech. Your words: " in the end there are significant, long-dwell systemic issues with raw material, industrial capacity, and workforce that will slow things down unless they are addressed. Iron, coke, chromium, and steel—and the people that put them on rolling stock to shipyards and factories—are suffering under the second-order effects of post-industrialization and low demand." Truer words seldom spoken. I'm gun shy of "transformation". Worked "joint experimentation" for a number of years, it failed, and I've got the receipts. Was victimized to put together the history of ACOM / US JFCOM in 2011. Six months of my life I'll never get back. If we approach "transformation" the way JFCOM did, it's time to start learning to speak Mandarin.
I wish the Secretary of War well but he is tilting at windmills.
We are good at telling ourselves what needs to be done. However, Clausewitz reminds us that commanders have a finite amount of will. "He recognized that real-world wars are limited by politics and resources, meaning "will" (resolve, desire, spirit) is finite and varies, leading to a mutual escalation until the cost of effort outweighs the political goal." I'm told that President Reagan kept an index card in his pocket with the three things that mattered (to which he applied his will). Someone else would have to worry about everything else. So, whose index card is this written on?
It would be helpful to know who wrote the speech, whether a speechwriter or an actual policy person with some imagination to suggest how to see change through. Hegseth does not possess the intellectual acuity to draft a serious document. The suggestion that a passed over National Guard major has morphed into a 21st century Clausewitz of strategic strength beggars the imagination.
WRT to Sal's question:
Wait…what? Someone needs to brief me on what was created that required a pushback against a “woke” and “social justice” AI. There is a backstory here I need to see.
A case of GIGO from what I can surmise. There are several famous cases of AI requests that returned unexpected results. Such as a request for "A picture of a WWII German soldier" getting a fairly accurately uniformed and armed soldier who was clearly of black African lineage. In essence the AI algorithms had been brought up woke so leaned to non-white like a TV commercial.
That was among the more egregiously stupid ones, but apparently text responses often ignored clear historical data and returned a woke slant to many things. Essentially, the database was corrupted with sources that had a slant to them.
Wish I had good sources to cite, but I tend to studiously avoid AI. Sort of feels like consorting with the devil to me. I'm sure a search will turn up some more on the subject.
AI did not even understood that there are some "historical data". AI is a statistical-based, pattern-seeking system. Its logic is not linear, cause-and-effect one, that humans have. Its logic is based on generalized statistical patters.
For example: for humans, 2+2=4 is logical, because if you took two apples and add two more (cause) the effect would be four apples. For AI, it would merely be "most likely 4", because statistically 4 is the answer most likely to be correct.
That's the reason for black German soldier in AI generation. AI have generalized soldier, and generalized German uniform. Generatling the image, the AI essentially generate a big field of random pixels, and seek in this randomness statistical similarities that correspond with the prompt ("A picture of a WWII German soldier"). Problem is, AI did not actually comprehend the casuistic interconnections here; it have no reason to seek specifically white Caucasian (yes, this term is outdated... but the whole racial theory is currently a mess).
Well, I get the lack of understanding of history and application of logic in AI*, but I think that this goes further than that. Certainly we can agree that historically the majority of soldiers were not black. That applies certainly to Europe and Asia, naturally much less so to Africa. In North and South America it is less applicable as well. Not sure about the demographics of South America, but certainly in North America the portion of soldiers that were/are black is significantly less than white though maybe equal with some other minorities.
The point of that being somewhere in the database the system is taught that soldiers are typically black when they are not. I suspect this is in large part due to the anti-war propaganda from the 60s on that says something to the effect of "black people are forced to fight and die in disproportionate numbers here white people avoid combat and the military all together (with an added caveat in the last decade "because of white privilege.)"
So still GIGO. The data given is selective. It would be interesting to see if an AI could actually manage a database of something close to all the world's knowledge.
Which raises yet another interesting point: Can AI database compilers add copywritten material without express permission? And what are the ramifications of having, or not having, such things in the database? While copyright covers a lot of ground something as simple as a book on a historical subject might be denied to the database when the contribution to gathering correct answers could be significant.
*Though if taught logic, I think that could be done as well...though applying if/then, and/or/nor logic to machine learning is likely very difficult in any but micro cases as the means of resolving takes some value judgements which is but the flip of a byte in a computer.
I like the AI search about dictators that returned Trump 47.
The thing about AI is, it's the right tool for a job, somewhere, but not for every job. Particularly, the following is not a use case for AI: getting factual answers to simple questions. AI doesn't do facts. It takes this or that piece out of its training data. If the training data is crap, the answer will be crap. If the training data is excellent, AI may still screw it up. The kind of AI we're all using now (large language models) has no concept of truth vs. falsehood, and cannot have that concept. Can AI see trends in data? Sure. Can it select a reasonable course of action from a constrained set of possible actions? Probably, but that should be as a decision aid to a human, not in place of the human. Part of AI research needs to be figuring out something useful we can do with it.
It seems like one of the most important things about how one uses an AI answer is understanding the database that it draws from.
However, due to the vast amounts of information that may be in such databases it seems likely that it would be hard for any individual user to truly know and understand the knowledge base.
One would have to take the designer's word/notes on it for what was included and the opinions that it would follow. I personally would be leery of accepting assurances from anyone I was not familiar with unless I did some checking myself. Cf.
Well, considering how the first couple years of the $1.5 trillion defense budget will have to go to infrastructure (nitrocelluose/gun cotton, artillery shell/missile production lines, shipbuilding/repair facilities), plenty of steel to be brought from the US metals industry.
"We killed the old defense acquisition system and created the warfighting acquisition system to focus on speed, risk and accountability. With the AI strategy and innovation ecosystem transformation that we have just outlined, we are welding that third piece into place."
....this 'old' guy is just under 18 Mondays from departing the 'bidness' for good BUT, while I'm still lending a hand, I haven't seen anything at the program office level indicating JCIDS is dead.....since my entry into this mess in 1983, "transformation" and the search for "speed to the Fleet" and other associated phrases haven't yielded what I'd call increased speed of product delivery.....I'm sure my cynicism is shining thru in this comment but let's not forget Newton's First Law (Inertia) - it is real and compounds with the 'this too shall pass' mentality (those acolytes raised in and nurtured by the system tend to dig their heels in)......in some ways, I'm kind of sad to be retiring as the coming three years - maybe 7 if the republicans can hang on - might be interesting within the five sided wind tunnel and the SYSCOMS.......
"There is another gem that demands a backstory, “data hoarding?”" Oh come on Sal, I'm sure you've seen it, Gods know I have. "Oh the Airforce did a study about that awhile back, but we can't seem to get the results."
"Yeah John's the only guy that knows anything about that system, you have to talk to him."
Between the guys that decide "knowledge is power, if I share the knowledge I lose power" and the guys that use knowledge for gatekeeping purposes, there's a lot of info hidden in desks and minds. Places where it helps an individual, but not the community. And it exists from the wrench turner on the waterfront, to the SES in his lofty office, 'far above the dirt and sound of work.'
I think that we may require some instructive executions.
Hegseth's CTO is a JD with a government BA!
The problem is there ain't enough STEM in DoW!
I left honest service in the operating AF 41 years ago and began a sentence in acquisition. t that time "Packard" commission was doing the initiatives for reforming.....
I wonder if Grok will thunder through a program's "earned value" reports and recommend a system be terminated when its estimate at complete is 130% with less than 30% value earned.....?
That is if Grok gets data that is not lies.
In my time I saw more programs that needed cancelled than US could afford. They were not canceled.
First, cancel dogs, then talk about why the system failed and recommend fixes.
Often good engineering technique was ignored and tossed aside because it was not required by law.
Can Hegseth get the toilets working on USS Ford?
LOL! Thanks, Ed. Let's see if Grok can tell us that the $400 hammer really cost $10 and the permitted overhead was $390!
Your $400 hammer is built to DoD specs.(prob. heavier than the hammer itself), bought by the Service AFTER the production run was finished and they bought ONE. In addition the Service in question bought NO spares or low-balled the buy during production(based on a flawed model of failure and or usage of said hammer). Yes, there is overhead, but cost is trapped in the Pentagon glacier known as the "procurement system". Reform is needed, program managers need to be terminated and Services can babble about STEM, but its time to put LOGIC(istics) back into procurement.
I remember the Packard Commission when I first came to DC in the 1980s. It want the fist or last attempt at reform.
As I pointed out in yesterday's comment, SECNAVs over the pass quarter century have not suffered from a lack of gold sleeved flag rank or STEM accreditation, yet here is where all of that "expereince" and insider knowledge has led us.
The UK started forcing mergers of aircraft companies about 75 years ago. The result was a few bloated giants that were too slow and cumbersome to respond to an overly bureaucratic and redundant procurement system. Twenty years later with the notable exception of the BAE Hawk they could not produce an aircraft without a foreign partner.
We started down that road with the last supper. There's might be too much inertia overcome this.
On the other hand, French post-war aircraft industry merged just fine. Granted, the merge actually started pre-war with nationalization of aviation industry.
The French pre-war nationalization was an abject disaster. Post war, I believe many companies were privatized and then later merged. However, Bloch (later Dassault) still remains private.
Thanks for the great and concise synopsis, Sal, and good to read you're cautiously optimistic about it in its totality. (One caution, though: please hyperlink to a different source than Wikipedia for background info or bios on people. Wikipedia is complete trash). Keep up the great work!
"An optimist believes we live in the best of all possible worlds,...and the pessimist believes that is true. (someone's quote)" Regarding "transformation" and AI, we've seen this movie before, and we've gained no superior real advantage. After advising a recent commission charged with "reforming" the Pentagon's end-to-end resourcing/budgeting systems ("acquisition system") on problems and potential solutions, their report failed to address those areas. I wrote my own small super critical report stating that none of their 28 recommendations would result in any substantial improvements (in an administrative process supporting operational requirements) and sent it to the four congressional defense committees with 14 alternate recommendations. But lead out with my own experience: "I like to kid I only learned two things in grad school. Lesson One: Hire your superiors. (Which means, even if you know how to do a job, when you need something done, go find the person best qualified and experienced and hire them.) Lesson Two: When all else fails, reorganize (...because it has the appearance of progress)."
Reading the CDR's comments this morning I'm still a bit if a pessimist regarding all this chatter about reform and AI. Especially the part about AI sorting out all this unshared data. The entire contracting system doesn't improve because WE don't really want it to change. Yes, some improvement, but we can learn faster all the dumb stuff. We really don't need AI to tell us we need to field more munitions (even on expendable commercial type ships) to take out all the navy ships China is building? Or that we need to produce drones at scale (the Ukrainian army is already teaching that lesson to the Pentagon in real time)?
One of my alternate recommendations: Let industry self-finance ten-year contracts and produce efficiently at scale, with DoD buying it out at delivery. Keep DoD/Navy out of the shipyards. Start with MSC. Recapitalize the entire supporting fleet now. No more small contracts for one or two boats. (If anyone wants to see my alternate recommendations on really improving the PPBE process, send me an email at johnoking82@gmail.com.)
One might begin to experience a slight glimmer of optimism.
One also hopes that it's not just a fever dream....
"game changing technology,"
This, I do not like. Someone needs to give Hegseth a copy of Arthur C. Clarke's "Superiority."
Sometimes, you get as much or more bang for the buck by improving your existing systems as you do from trying to build shiny new ones.
(See: the T-34/76 to the T-34/85 and the numerous Sherman variants, especially the M4A3(76)(W), as opposed to the German Maus.)
The US must be able to make the raw material, process it, and lay down hulls all in-nation.
Otherwise, its over.