"Why are you still alive?"

Theemile wrote:Given what happened to Galton, Darius might have been found by Jesssk's exploration ships and then downplayed in the database, or the data was altered and ship killed like Galton. who knows.

It's possible they faked a survey just to insert a "nothing to see here, move along" datum into the database. There's no way to really know until RFC tells us.

Brigade XO wrote:If Darius is at a star just the other side of the Felix wormhole, then it is reasonable to guess that it was possibly the nearest and 1st star listed by the ship that either did the original investigation or the 2nd ship---one that was sent thought to see what was around that terminus if the original prospecting ship didn't do more than gather it's terminus data and go on the return trip as soon as they were confident of making the return trip.

It's possible, but there's no evidence that Darius is the first terminus discovered through the Junction. It could be any of the four.

In any case, that's a completely useless piece of information. The ship that did the surveying never reported anything outside the MAlign. Either it was a MAlign ship, legitimately paid for through a shell company like Jessyk, or stolen for the purpose (hired but never allowed to return). Even the Mannerheim forces weren't told that there was a junction in their backyard for 175 years.

If the explorer ship is working for the Alignment, nobody else will get any of the information....ever.

tlb wrote:Speaking of the spider ships; if Gail is a security threat, then what about all the crews of the Ghost and Shark ships that participated in Oyster Bay and their families that might have figured out what they did? Come to think of it, as a weapons analyst, Gail must have known about Oyster Bay and most likely the Silver Bullets at Beowulf (perhaps even the bombs in the orbitals). So why is she suddenly a security risk over the defense of Galton, that is the least of what she must know.

It's not the offensive operations that are the problem.

Rather, she knows Galton was sacrificed. (She doesn't know the name, though.)

ThinksMarkedly wrote:Which means Darius might be accidentally found by survey ships now coming near it or by an unknown colony ship that was going for it all along and has been in transit for more than 2 centuries.

That's one event that cannot be predicted by either side, nor by us. It could be a way for RFC to give the GA the necessary data.

Disagree. I have previously said they need enough of a local fleet to ensure any pirate that happens by doesn't live to tell the tale. Scouts would likewise have to be eliminated.

There's no colony ship headed there--since there is some system to keep multiple groups from attempting to colonize the same planet there must be some record of where colony ships are heading. That would have been checked before picking the planet.

tlb wrote:There is another point of interest about her: why would she be considered a security risk? It is not just that she knows about "System Alpha", but that she actively worked on improving their defense.

Loren Pechtel wrote:She will realize it was a real system that was lost.

tlb wrote:It seems to me that Gail is only a security risk to that cover story; but there are too many inconsistencies for that story to work, if and when Darius is found.

Loren Pechtel wrote:No, this is the light that fell from the sky on The Truman Show. Inconsequential by itself, very relevant in that it will make her realize society isn't as everyone believes.

tlb wrote:Speaking of the spider ships; if Gail is a security threat, then what about all the crews of the Ghost and Shark ships that participated in Oyster Bay and their families that might have figured out what they did? Come to think of it, as a weapons analyst, Gail must have known about Oyster Bay and most likely the Silver Bullets at Beowulf (perhaps even the bombs in the orbitals). So why is she suddenly a security risk over the defense of Galton, that is the least of what she must know.

Loren Pechtel wrote:It's not the offensive operations that are the problem.

Rather, she knows Galton was sacrificed. (She doesn't know the name, though.)

You have said this before (and I have included those quotes), but I do not understand the point that you are trying to make. Gail is not an innocent, living a life that she does not know is staged; as in the "Truman Show". She is a member of the Onion and a weapons analyst; so she knows bad things about the Malign, among them is knowledge of the weapons and performance of the missiles used in Oyster Bay and at Beowulf. If she is disillusioned at all, it is because she wanted to use those same weapons at "System Alpha" and was denied.

So, what is it about knowing that Malign lives were lost (even though she knew the defense was doomed, no matter what weapons were used), that is worse than the knowledge she has of all the deaths that the Malign has caused?

cthia wrote:

Jonathan_S wrote:But even if we say the first case is true I wouldn't give a such a model much credit for predicting Honor's next targets; not when the best they seem to be able to say about it is that it is not "a good predictive model".

I agree. As you said I think the author wanted to infer that. But, I hardly think Linda and Lewis was willing to leave the entire thing up to fortune telling without having consulted an expert system. The expert system was simply unclear or incomplete. I wouldn't be surprised if the predictive model would have come up with the solution had they had more time refining the data and asking it the right questions. They just didn't have the time. They had enough data. Because the data that their intuition worked on (the tea leaves) existed.
[snip]

I agree that RFC was trying to convey that it was human intuition. But purely intuition? Without access to any of the known data? Intuition works on data, even if subliminally. Intuition isn't "conjured up" out of thin air.

Jonathan_S wrote:"Purely intuition"? Of course not.
That's why I said "it was in large part human intuition that led them to winnow down the entire Republic to a list of 10 most likely, and 15 more still likely, target systems that Honor might try for.". "in large part" != "purely".

Both you and I used essentially the same qualifier. Your "in large part" and my "for the most part." So we do seem to agree there.

Jonathan_S wrote:However I'm less convinced they had enough data for the predictive model to work with if only they'd had more time. It seems to me that the issue is identifying which system characteristics the Manties were looking at, and then how they were weighted.

Absolutely. And I think they had enough data already.

Jonathan_S wrote:However maybe I'm making a different distinction, or using terminology slightly differently, than you are. I'd view the expert system as the software that attempts to develop a predictive model based on training datasets and other input given to it, and the predictive model as the result of feeding the expert system such data.

We are in agreement on the terminology as well. To be concise, an expert system is really just the generic off the shelf software that can be used to create an expert system. Let me add that a couple of terms are being used interchangeably in today's markets that really are misleading. To be more accurate, an expert system is really just the off the shelf generic software that is used to create a specific kind of "knowledge base." In today's markets they claim that expert systems are the forerunner of a knowledge base. But an expert system doesn't become an expert system until you feed it a specific kind of knowledge base. For example, my sister surprised me over twenty years ago when she told me she uses an expert system in the medical field.

You can train the generic software package to become an expert in any field by feeding it loads of data. That is its knowledge base. It doesn't become an expert until you feed it.

Haven would have had an expert system developed for, what, decades? Centuries? That is a lot of knowledge being fed to it, and constantly being updated on its systems. After all, as I said, I imagine the Hollow Tank has access to the very same knowledge base fueling the Admiralty's decisions about fleet dispersions, pickets, etc. But I imagine the analysts and tea leaf readers access it from different stations to keep the Hollow Tank clear for the Admiralty.

Jonathan_S wrote:Yes the Republic Navy's intelligence analysist would have had expert systems to try to develop predictive models -- but in this case they didn't feel there was enough data to let the expert system come up with a good model.

You are correct. Lewis said that be didn't think they had enough data. I disagreed with that in another post. They did have enough data. Also as I mentioned, I am surprised that he would even suggest that. It is impossible to know if you have enough data until you submit the data to be modeled. Because even though the data that you submit may be sparse, the knowledge base compares and sorts it to a humongous mound of information. I have seen expert systems spit out a conclusive answer from an insanely small input.

Jonathan'S wrote:(But I don't know whether they made the decision by trying and seeing it gave poor results; or whether they knew by past experience how much data would be required and knew they didn't yet have that much data)

As I said upstream, it is impossible to know if you have enough data until you query the expert system. The expert system may have been able to produce an adequate model after just Cutworm I. That was only one operation, but it was comprised of five different targets.

Jonathan_S wrote:There had only been 2 series of raids before they were successful in ambushing Honor at Solon. And put together Cutworm I & II had hit only 9 systems out of the entire Republic. That's such a small data set that even if you'd perfectly identified the key characteristics the Manties were using for target selection there would be multiple different relatively weightings of those characteristics which would all give you that same 9 target set -- but which would result in quite different guesses as to the most likely next target(s).

First off. Nine systems translate into a lot of data invoked by the expert system's knowledge base, that has been compiled on those nine systems. As I said above from my long experience with expert systems, the five targets hit by Cutworm I should have given them a very promising model, if not a good one. Consider that their expert system has a knowledge base going back many decades or centuries. So, just the five systems that were hit in Cutworm I would have triggered a lot of data in the knowledge base.

Jonathan_S wrote:Further, the initial report appears to have predated Cutworm II, since Marquette and Theisman were discussion the success at predicting Des Moines, Fordyce, and Chantilly -- all of which were hit in the 2nd set of raids. So the initial report only had the 5 targeted systems from Cutworm I to use as data points to try to work out where Honor's forces might strike next. From those 5 they worked out 10 primary targets (2 of which were correct) and 15 less likely targets (1 of which was correct).

So I was somewhat wrong before. They had called the model "not good" after placing their bets. But I'd missed that it was their bets for Cutworm II. With the additional data points of the 4 systems hit in that 2nd phase of raids they would have been able to try to refine the predictive model and it might well have been much better (possibly even considered good) before they had to make their (ultimately successful) bet on the target(s) for what turned out to be Cutworm III.

Yes! As I said, I would have been surprised if the expert system didn't yield promising results after just Cutworm I. Cutworm I was five systems. There is an awful lot of data stored about those five systems compiled and collated for centuries. With the additional data on four more systems and still no good model really surprises me.

The expert system's knowledge base would include everything that fueled the tea leaf readers' intuition. Things like ...

"Basically," Marquette said, sitting obediently, "they tried looking at the problem through Manty eyes. They figure the Manties are looking for targets they can anticipate will be fairly lightly defended, but which have enough population and representation to generate a lot of political pressure. They're also hitting systems with a civilian economy which may not be contributing very much to the war effort, but which is large enough to require the federal government to undertake a substantial diversion of emergency assistance when it's destroyed. And it's also pretty clear that they want to impress us with their aggressiveness. That's why they're operating so deep. Well, that and because the deeper they get, the further away from the 'frontline' systems, the less likely we are to have heavy defensive forces in position to intercept them. So that means we should be looking at deep penetration targets, not frontier raids."

Even your very own criteria in a post upstream is really quite good ...

Jonathan_S wrote:Maybe. But I think the most useful thing a computer could do in this situation is be a queryable database of all kinds of pertinent information on systems and their planets.
Letting the annalists think up various parameters or characteristics (location, industry, political clout, defenses, etc.) and get back the resulting lists of matching systems. (Because there's too many systems to have them all in your head)

The problem with using a generic computer is that it will simply provide you with raw data, of which there would have been tons of uncollated, cross referenced and analyzed for the very patterns that you are looking for. You would essentially end up with a lot of data on your desk like in the old days with many eyes looking for that certain information found on two separate pieces of paper that form a clue. An expert system does the filing and collating and grouping together of facts. Even just Cutworm I should have generated tons of information. An expert system will present that information in an efficient form that is easy on the eyes and brain. Like a spreadsheet arranges statistics.

Jonathan_s wrote:"Okay, if we assume 8th fleet doesn't want to hit anything with more than 6 weeks travel out; but has at least X reps in parliament, what systems does that leave? Okay, let's try filtering out the ones with fleet bases as being too tough a target; now what do we have?"
"That looks promising, but lets see if anything significantly better pops out if we push the range out to 8 weeks"
"I think they'll weight naval infrastructure damage more highly, which systems have repair yards or missile depots?"
"No, I think they'll want to take out the most mobile defenders that they safely can; rank the systems by size of their modern defensive task forces"

Very good criteria.

Jonathan_S wrote:That's not going to give you your final answers; but its a quick way to filter out the majority of the chaff based on various criteria. Then you get different analysists all applying their own theories on Honor's targeting criteria and sit down and compare lists with each other.

It didn't give the final answer, but it just as easily could have. At any rate, you wouldn't want to make the job more difficult for the analysts with simple raw, un crossreferenced, uncollated data that hasn't even been checked for patterns in a way that only an expert system can. An expert system is an expert on patterns and they were looking for patterns. The final submission was itself a pattern conceived by the readers. From the same data that should have long been a part of the knowledge base.

Jonathan_S wrote:I doubt an expert system would be of much use because what they're struggling for is gaining intuition on how Honor and the Admiralty are selecting their targets -- so humans thinking about possible alternate criteria seems the most important part. Letting an expert system do that "thinking" for you kind of defeats the purpose -- at least when you don't have enough training data for it to get good. In this case figuring out whether the judgement criteria makes sense, in the absence of sufficient data, is the key thing -- and that's the thing expert systems are worst at; because they can't apply intuition to insufficient data.

But there wasn't insufficient data. That is where we disagree. The expert system should have had the same data set that fueled the tea leaf readers' intuition. The same tea leaves that I posted above that are included in the textev.

Not just sometimes, or most of the time, but all of the time the success of an expert system will always depend upon asking the right questions, even if you have a complete data set to present it.

This reminds me of the many times Geordi La Forge and Commander Data had gotten annoyed while querying the Enterprise's knowledge base on a certain matter. They caught no joy until they asked the right question with the right command. The most memorable case was when Dr. Crusher was querying the data base trying to determine where the rest of the crew of the entire ship had gone. Why was she the only one left on the Enterprise. "Oh shucks! The answer has to be there. Well computer, give me the answer to this. And cross reference it with that." "Bingo!"

Jonathan_S wrote:Actually generating your final selection is just a result of finally converging on a collective best guess of how your enemy seems most likely to pick their targets.

An educated best guess implied from inference.

Jonathan_S wrote:I don't think we know how many systems were on the more likely and less likely lists for that; but they obviously included Solon which was one of the 2 systems Honor's forces did hit. (Though why her other target, Lorn, didn't have an ambush of its own set isn't, to the best of my recollection, explained. Could be the analysists missed that one, or rated it too low on their list; or could be the RHN had other reasons for which systems they could afford to set their few concentrated forces at.)

I don't have searchable text, but I do recall that Honor actually had a method to her madness. Part of it was the pressure that would be put on their navy from a political standpoint. There were at least several considerations as I recall. So, Honor's own model should have been a subset of the expert system's knowledge base.

Jonathan_S wrote:I don't think we know how many systems were on the more likely and less likely lists for that; but they obviously included Solon which was one of the 2 systems Honor's forces did hit. (Though why her other target, Lorn, didn't have an ambush of its own set isn't, to the best of my recollection, explained. Could be the analysists missed that one, or rated it too low on their list; or could be the RHN had other reasons for which systems they could afford to set their few concentrated forces at.)

In order to concentrate force to punish Honor, they only set up a trap at one of the worlds expected to be hit, rather than putting smaller forces at several potential targets.

cthia wrote:I don't have searchable text, but I do recall that Honor actually had a method to her madness. Part of it was the pressure that would be put on their navy from a political standpoint. There were at least several considerations as I recall. So, Honor's own model should have been a subset of the expert system's knowledge base.

Yes; if the analysts were good, then Honor's criteria would have been duplicated in the parameters used to judge where to place the reaction force. But your phrasing make it sound as though Honor's model was openly available as an input.

Loren Pechtel wrote:Disagree. I have previously said they need enough of a local fleet to ensure any pirate that happens by doesn't live to tell the tale. Scouts would likewise have to be eliminated.

There's no colony ship headed there--since there is some system to keep multiple groups from attempting to colonize the same planet there must be some record of where colony ships are heading. That would have been checked before picking the planet.

Fair enough, that should take care of anyone who drops by to say hello. They won't be coming back to report they saw anyone. So for a survey ship, the question is whether they drop in close enough to a defensive installation that could intercept them before they leave and if they decide to leave before seeing who's there. That is, if they drop out a few light-minutes from the hyperlimit and notice there's a huge industry around, do they think "let's go say hello" or "this is suspicious, let's go report it?"

There can be colony ships going for Darius because not all all colony expeditions would have been registered. There can certainly be unauthorised ones launched recently from the Fringe, or an older sublight one that never bothered to send an update to Earth. Or, for that matter, one that did send the update but which never arrived: if they were going sublight, it's because hyperspace was dangerous, so the it stands to reason that an FTL ship with the update could have been lost.

Anyway, I'll agree the chance that an accidental discovery is the reveal is minimal.

cthia wrote:You are correct. Lewis said that be didn't think they had enough data. I disagreed with that in another post.

In other words, you're arguing with the character. We are taking the character's statement at face value because we have no reason to suspect him of lying or dissimulating.

As I said upstream, it is impossible to know if you have enough data until you query the expert system. The expert system may have been able to produce an adequate model after just Cutworm I. That was only one operation, but it was comprised of five different targets.

First off. Nine systems translate into a lot of data invoked by the expert system's knowledge base, that has been compiled on those nine systems. As I said above from my long experience with expert systems, the five targets hit by Cutworm I should have given them a very promising model, if not a good one. Consider that their expert system has a knowledge base going back many decades or centuries. So, just the five systems that were hit in Cutworm I would have triggered a lot of data in the knowledge base.

They're also trying to predict a set of changing conditions. Cutworm I was one operation, then the Alliance escalated with Cutworm II and the parameters for selecting the target changed. The expert system needs to predict when those parameters would change to make the next prediction. And once that change has happened, the expert system needs to reasonably well separate the two data sets so one doesn't pollute the other with points outside the curve.

And as I said in my previous post in this sub-thread, who's to say that there wasn't an actual portion of randomness in the Alliance target selection? The Alliance does have access to the same software, if not better, and if they could predict their own targets using the same or better data set, they should assume the RHN could too. But you can't predict random.

Yes! As I said, I would have been surprised if the expert system didn't yield promising results after just Cutworm I. Cutworm I was five systems. There is an awful lot of data stored about those five systems compiled and collated for centuries. With the additional data on four more systems and still no good model really surprises me.

Promising results, yes. It did generate that. The problem is that there were still too many options, so they couldn't all be well-defended.

This reminds me of the many times Geordi La Forge and Commander Data had gotten annoyed while querying the Enterprise's knowledge base on a certain matter. They caught no joy until they asked the right question with the right command. The most memorable case was when Dr. Crusher was querying the data base trying to determine where the rest of the crew of the entire ship had gone. Why was she the only one left on the Enterprise. "Oh shucks! The answer has to be there. Well computer, give me the answer to this. And cross reference it with that." "Bingo!"

TNG episode "Remember Me," the single weirdest logical conclusion, ever: "if there's nothing wrong with me, maybe there's something wrong with the Universe."

ThinksMarkedly wrote:And as I said in my previous post in this sub-thread, who's to say that there wasn't an actual portion of randomness in the Alliance target selection? The Alliance does have access to the same software, if not better, and if they could predict their own targets using the same or better data set, they should assume the RHN could too. But you can't predict random.
[cut]
TNG episode "Remember Me," the single weirdest logical conclusion, ever: "if there's nothing wrong with me, maybe there's something wrong with the Universe."

Another Star Trek reference comes to mind: Ferengi Rule of Acquisition 75:

Once in a while, declare peace. It confuses the hell out of your enemies.

The best way to make sure you can't predicted is to be unpredictable. It's not too difficult to insert noise into your signal so those expert systems can't be accurately used.