The company these guys started (Autonomy) and sold to HP was a sort of pioneer in applied Bayesian math. I was a user of it on a project I worked on around 2001. It was one of the first (maybe THE first) commercially available products that did searches and clustering analysis based on a probability assessment of digital signatures rather than keywords. So if you were searching a lot of text for dolphins, instead of doing keyword searches on dolphin, fish, ocean, swim, etc., you acquired a set of training documents with info about dolphins, and the software would analyze a data set and match digital signatures against the signature of dolphins, using Bayesian math, to find documents about dolphins.
Sorry for the geeky explanation, but the point is that this boat was named Bayesian for a very good reason. My perspective might be off, but I think Autonomy's application of the Bayesian method was a very early pioneer of something that is outright ubiquitous in IT today. Would also explain why they sold it for $11 billion dollars, and perhaps why they were acquitted of accusations that that price was illegally inflated.
Thank you very much for this explanation! I've just finished reading about the various allegations of fraud associated with the sale of Autonomy to HP and the various legal entanglements which followed. I have a hard time understanding tech stuff and I have little understanding of "signature," but I have an impression of pre-data or that which might construct the searched-for item and then narrowing the identifying tidbits or "signature" to that deep-based search matching up with other identifying (at this elemental or deep level) bits or other "signatures." Indeed, such innovation would be enormously powerful in the digital world. The HP CEO at the time of purchase was greatly enamored with Autonomy's potential despite acknowledging an over-price. I think HP shareholders weren't and that, perhaps, HP's financial position could ill afford the absorption of such a highly speculative venture.
...Or. HP sold, when the second CEO balkanized Autonomy, that deep-seeking search process to DARPA. Still, they would have been well compensated so we are still left with cui bono? Or who didn't benefit enough financially.
I believe the legal charges rested with the alleged overvaluing of the Autonomy's sale's price. Viewing the current DOJ's many attacks against Trump on specious charges, I noted a like attack against Lynch while reading the background. The original CEO of HP when Autonomy was bought was accused of buying over price value and was warned in a pre-purchase report, and another CEO, who appeared to have political ambitions, was brought in who began dismantling Autonomy.
Evidently, shareholders initiated much of the trouble as they fervently opposed the purchase at that price. The entire episode looks highly politicized as in "company politics." One financial officer was convicted in 2018 and spent five years in jail, which I couldn't tell if the sentence was here in the U.S. or in the UK where several suits unfolded, too.
Money...it's money that caused this problem...only I can't figure out why. Who were the shareholders or stockholders in HP at the time? Or who, in actual fact, controlled those shareholders? And for what purpose? Just as Vanguard's BOD is never listed (and Vanguard essentially owns everything when stock ownership is traced through companies), I wonder what high-level money scheme was thwarted or made negligible.
P.S. If I understand what you wrote about Autonomy's software, this software could easily be used for "pre-crime." I.e., the terror of "predicting" who will "commit" crime.
Thanks for the case summary. I haven't looked into it, and that's some helpful background. One of the challenges in even determining a motive if the men were murdered is that it could have been for financial gain or just for spite. If I were Columbo and working this case, I'd be talking to all the disgruntled HP shareholders. But I guess that episode would run a few decades. lol
I'm not an expert on the technology behind autonomy. But I think your supposition that it could be used for pre-crime is probably on-target. You can think of digital signatures in this context like this: Autonomy has a digital signature for all the concepts it's been trained on. And you can think of each signature as a set of parameter values, say a =100, b = 12, c = 75, which of course is an oversimplification. And it calculates a similar signature for each "document" in its database. Then it uses algorithms based on Beyesian math to calculate a score that expresses how close the signature of each document is to the signature for the concept you're searching on.
At the time when I was using Autonomy, a unique feature was the ability to search video and audio files using the same concept. You could feed it an image, for example, and it could find pictures or videos in its database which the Bayesian math said were matches to that image you were searching for. Same with audio files.
Of course, everyone and their dog is making apps now that do something like this, Google's image search is a well known example, but I imagine there are many thousands of examples. I don't know to what extent or in what manner Autonomy is upstream from these apps, but clearly it was an early application of a technology that is truly ubiquitous now.
The irony is that the owners died in an improbable way in several respects. I wasnтАЩt commenting on the product produced by the innovative use of this math concept.
If nothing else, that may be the most ironic boat moniker ever.
The company these guys started (Autonomy) and sold to HP was a sort of pioneer in applied Bayesian math. I was a user of it on a project I worked on around 2001. It was one of the first (maybe THE first) commercially available products that did searches and clustering analysis based on a probability assessment of digital signatures rather than keywords. So if you were searching a lot of text for dolphins, instead of doing keyword searches on dolphin, fish, ocean, swim, etc., you acquired a set of training documents with info about dolphins, and the software would analyze a data set and match digital signatures against the signature of dolphins, using Bayesian math, to find documents about dolphins.
Sorry for the geeky explanation, but the point is that this boat was named Bayesian for a very good reason. My perspective might be off, but I think Autonomy's application of the Bayesian method was a very early pioneer of something that is outright ubiquitous in IT today. Would also explain why they sold it for $11 billion dollars, and perhaps why they were acquitted of accusations that that price was illegally inflated.
Thank you very much for this explanation! I've just finished reading about the various allegations of fraud associated with the sale of Autonomy to HP and the various legal entanglements which followed. I have a hard time understanding tech stuff and I have little understanding of "signature," but I have an impression of pre-data or that which might construct the searched-for item and then narrowing the identifying tidbits or "signature" to that deep-based search matching up with other identifying (at this elemental or deep level) bits or other "signatures." Indeed, such innovation would be enormously powerful in the digital world. The HP CEO at the time of purchase was greatly enamored with Autonomy's potential despite acknowledging an over-price. I think HP shareholders weren't and that, perhaps, HP's financial position could ill afford the absorption of such a highly speculative venture.
...Or. HP sold, when the second CEO balkanized Autonomy, that deep-seeking search process to DARPA. Still, they would have been well compensated so we are still left with cui bono? Or who didn't benefit enough financially.
I believe the legal charges rested with the alleged overvaluing of the Autonomy's sale's price. Viewing the current DOJ's many attacks against Trump on specious charges, I noted a like attack against Lynch while reading the background. The original CEO of HP when Autonomy was bought was accused of buying over price value and was warned in a pre-purchase report, and another CEO, who appeared to have political ambitions, was brought in who began dismantling Autonomy.
Evidently, shareholders initiated much of the trouble as they fervently opposed the purchase at that price. The entire episode looks highly politicized as in "company politics." One financial officer was convicted in 2018 and spent five years in jail, which I couldn't tell if the sentence was here in the U.S. or in the UK where several suits unfolded, too.
Money...it's money that caused this problem...only I can't figure out why. Who were the shareholders or stockholders in HP at the time? Or who, in actual fact, controlled those shareholders? And for what purpose? Just as Vanguard's BOD is never listed (and Vanguard essentially owns everything when stock ownership is traced through companies), I wonder what high-level money scheme was thwarted or made negligible.
P.S. If I understand what you wrote about Autonomy's software, this software could easily be used for "pre-crime." I.e., the terror of "predicting" who will "commit" crime.
Thanks for the case summary. I haven't looked into it, and that's some helpful background. One of the challenges in even determining a motive if the men were murdered is that it could have been for financial gain or just for spite. If I were Columbo and working this case, I'd be talking to all the disgruntled HP shareholders. But I guess that episode would run a few decades. lol
I'm not an expert on the technology behind autonomy. But I think your supposition that it could be used for pre-crime is probably on-target. You can think of digital signatures in this context like this: Autonomy has a digital signature for all the concepts it's been trained on. And you can think of each signature as a set of parameter values, say a =100, b = 12, c = 75, which of course is an oversimplification. And it calculates a similar signature for each "document" in its database. Then it uses algorithms based on Beyesian math to calculate a score that expresses how close the signature of each document is to the signature for the concept you're searching on.
At the time when I was using Autonomy, a unique feature was the ability to search video and audio files using the same concept. You could feed it an image, for example, and it could find pictures or videos in its database which the Bayesian math said were matches to that image you were searching for. Same with audio files.
Of course, everyone and their dog is making apps now that do something like this, Google's image search is a well known example, but I imagine there are many thousands of examples. I don't know to what extent or in what manner Autonomy is upstream from these apps, but clearly it was an early application of a technology that is truly ubiquitous now.
The irony is that the owners died in an improbable way in several respects. I wasnтАЩt commenting on the product produced by the innovative use of this math concept.