Forgot your password?
typodupeerror

Comment: Re:soooooo... (Score 2, Informative) 77

by basil montreal (#24799287) Attached to: IBM Flash Memory Breaks 1 Million IOPS Barrier
MB/s is only a measure that's meaningful for sequential reads where the data can be prefetched. Most enterprise storage is based on applications that read randomly all over the disk (like databases and email servers). The benchmark measurement for this type of application is in the number of operations you can do per second. A single hard drive spindle can do between 80 and 150 IO/s, which would generate the number of IO/s times the size of the IO blocks per second.
The Internet

+ - Comcast Weans Hogs Off Their Packet Teat

Submitted by
eldavojohn
eldavojohn writes "The telecommunications giant Comcast has severed its services to internet hogs who use more bandwidth than others. From the article, "Carreiro said he received a message from a Comcast Security Assurance representative in December, who warned him that he was hogging too much of the company's bandwidth and needed to cut down. When Carreiro contacted customer service about the call, they had no idea what he was talking about and suggested it was a prank phone call. Unconvinced, Carreiro contacted Comcast several more times, but was again told there was no problem. A month later, he woke up to a dead Internet connection. Customer service directed him to the Security Assurance division, which Carreiro said informed him he would now be without service for one year." This is quite alarming to me, considering that I am forced into using a particular ISP based on some deal my neighborhood made many years before I moved here."
The Almighty Buck

+ - Tech Entrepreneurs Strike Back

Submitted by sleeper555
sleeper555 (1076835) writes "A new web site, TheFunded rates venture capitalists, giving entrepreneurs more information about whether or not to take money from a particular VC This is a very useful service to have, because the venture process is still secretive, and there is little public and objective review information about venture capital firms and their partners. TheFunded, lets its "members" rate a venture firm (using a scale of 0 to 5) on each of the following criteria: its track record, operating competence, pitching efficiency, favorable deal terms, and execution assistance. Notably, you have to become member to submit ratings and leave comments, and you can't be a member if you are partner or an agent of a firm. Also, you can only become a member if you are invited by another member or somehow convince TheFunded that you should be a member."
The Internet

+ - Death and the Internet

Submitted by
yorickwilks
yorickwilks writes ""The Internet changes everything" says William Dutton, Head of Oxford's Internet Institute, but surely not death, which, with taxes, tends to be permanent and unchanging?
A moment's reflection shows that is not so, and civilizations differ in nothing more than how they treat death and its subsequent state, if any. Even within a single society, our treatment of death, and our reactions to it, are so different from those of our Victorian ancestors — with their crepe bands, black-rimmed newspapers and mourning hats — that they now seem more like a foreign tribe.

Shrines are normally for the dead, whether elaborate tombs or small candles in the corner of Japanese living rooms; but Internet shrines are different, and it is often hard to know if the owner of a web page is alive or dead. I turned recently to the URL of an Italian I had once met, and discovered from his webpage he was now famous as the Father of Italian Cybernetics. I had no idea he had become so celebrated, but what was hard to discover was the answer to what had brought me there: whether he was alive or dead. It is this crucial borderline — dead vs. alive — critical to physiology and the law, that the Internet systematically fudges.

If, in the future, to be is to be in cyberspace, then an area ripe for development is email, where there are huge commercial opportunities for postmortem email that are well within technical reach, given the advances in computer processing of natural language in recent years. With only simple electronics, we are all familiar now with goodbye videos from the deceased at their own funerals, and some modern gravestones, known in America as Vidstones, have, instead of a simple memorial picture built into the stone, a small video of the deceased, solar-powered, that can be activated by a switch.

But email has been subjected to a wide range of analysis techniques to understand its content, at some acceptable level, most notoriously by companies and national security agencies: but such analysis can also decide whether the email content is positive or negative, whether it is asking for something or announcing a lecture, and so on. These and many other kinds of email content can now be extracted by reasonably intelligent programs, and it would be a very small move to have them automatically replied to as well, a service that could continue without a living correspondent. This would require not only the special software but a stable host server, such as a university or perhaps, in the future, a foundation specifically set up for the purpose of repling to emails to the dead. That would keep someone in cyberspace, as it were, for a fixed period for a fee, just as medieval chantries would pray for your soul for so many years for the appropriate donation. Even now graves are rented only for a fixed period.

One's email could move seamlessly into an after-life mode: academics, for example, are used to accepting and declining invitations to lecture by email, and sending out their publications as offprints as attachments in the same way. It is perfectly straightforward to extend the standard Unix "Vacation program" which normally replies to email by saying you are on holiday and when you are coming back, so as to say:

I am sorry I cannot take up your invitation to lecture at your University because I died on September 1st 2008. I would have loved to come and see you all again; thank you so much for asking me

And so on. Sending out a requested offprint electronically would be straightforward, as would the provision of bibliographic or autobiographical information with the standard search technologies Information Retrieval and Information Extraction.

A recent Internet development relevant to all this is Second Life (https://secondlife.com/), a virtual world where some two million people have taken up a form of residence using avatars: artificial appearances or simulations of themselves who meet others, including the avatars of people not currently on line (at the time of writing only eleven thousand of the two million subscribers are actually on line). Second life has obvious resonances of "after-life" as well as "parallel life", which is the one its creators intended. The expansion of Second Life is extraordinary, with, at the time of writing, sixty thousand acres "sold", a space which is (virtually) expanding by 8% a month. The sales are in "Linden dollars" inside Second Life but they can be bought and sold for real money elsewhere on the web, which has given the virtual economy aspects of a real one.

Famous singers are now releasing songs within Second Life which the buyer can get and play there. I had a perfectly serious conversation last month trying to convince the British Library — which sometimes doubts the quality of its "outreach' in a demotic age — that it could buy land in Second Life and erect at least a large hoarding on it, saying "LIBRARY HERE SOON".

But what has this to do with the Internet? Well, life as an avatar, after one's own death, would certainly be a form of life, even in a virtual world. Your avatar on the other side could continue to function and appear to meet people, talking as it had been programmed to, visiting places and living a full if rather thin second life. It would not be you, of course, and at best rather like the schoolmen saw the lives of angels, as all form but no substance.

Other possibilities arise from the opportunity we will have very soon of putting every possible fact, memory and datum about our lives onto the Internet itself. Last week the British library hosted a meeting on Memories for Life ((http://www.telegraph.co.uk/connected/main.jhtml? view=DETAILS&grid=&xml=/connected/2006/12/13/nlife 13.xml), a topic generated as a challenge for modern computing a few years ago and funded since then as a university research network (www.memories-for-life.org). The core idea is that computer storage is now so large and cheap that it becomes feasible to envisage recordings of everything we say, hear, write, see, eat or meet in eighty years of life, along with all our medical readings, as being stored on the Internet in some moderate and quite available space (known to the technically minded as 28 terabytes, according to an estimate by Alan Dix of Lancaster University). This is a huge amount of space but, at the present rate of progress, will be quite cheap to buy with a year or two, and could be stored on something no bigger than a sugar lump. The real problem is how we could possibly search all those memories and facts, even if we had them available; how we could begin to make sense of them, in the way a biographer makes sense of a subject's life by careful editing and selection.

The notion of all life on the Internet may be fanciful, but it is plain that huge chunks of our lives are going to be put there, not only the emails and the documents we write, but all our photographs and videos, which we are being encouraged to show to everyone in the world on sites like flicr (flicr.com) and YouTube (youtube.com). A possibility discussed at the British Library was that people may be able in the future to organise their lives and memories and data on the Internet with the aid of special automated assistants that can talk and converse in a normal way. This is the same motive as that behind all the recent offers by Google, Vodafone and others to offer free storage space to the public for their life data; but in return, the companies will get access to people's memories, tastes and records and, under certain safeguards, will know what adverts to send personally to them.

Something slightly more academic and benign is the notion of a computer Companion: a conversational agent that stays with one for a long period, appears to learn one's tastes and habits, and helps to organize and select all this personal material for its "owner". One can think of a Companion as best suited to the elderly, living alone and in need of company, needing to be reminded when to take pills and of the soap opera plots if they have been forgotten. Concretely, the Companion could be a mobile phone, or a computer screen, but more likely something like a furry handbag that sits on one's lap and talks, light to carry about and definitely not a robot. The Japanese have already gone some way in this direction: the BBC website carris a story of an elderly Japanese lady, Akino, who has a commercial companion called, Primo Puel, whose Japanese is primitive but Akino is reported as being comforted by it, liking to "hear it chatting away to itself in the other room". Akino added that she found more comfort in it than talking to her late husband's shrine in the corner. Readers will remember Japanese toys without any language like Tamagochi that had no language at all but aroused powerful emotions of care towards them in their owners.

Primo Puel has no language but natural computer conversation has come a long way even if it has not yet the everyday success of machine translation, which can now give a reasonable if basic translation of any web page on demand. The Loeber website (www.loebner-competition,org) shows the year by year increasing capacity of programs that take part in its annual computer conversation competition. There is now a major European initiative to build a computer Companion using these much stronger methods for deriving computer conversation: it is called Companions (www.nlp.shef.ac.uk/companions) and I coordinate it at the University of Sheffield, where it will run for four years with fifteen EU and US partners at a cost to tax payers of some 13 million.

It is said that the elderly in care homes spend much of their time shuffling their memories in the form of photographs; soon these will be digital images, of course, and the EU Companion will start there, discussing with its elderly owner who is in each picture, where it was taken and what its importance is. The idea, which may or may not succeed, is to use conversation with the elderly to build up coherent narratives, stories of parts of the owner's life, the stories which the images tell. Normally only those with talent, resources and leisure write autobiographies, but, if the Companions project succeeds, everyone could assemble some form of autobiography for their children, and undergo, with the Companion's help, some form of debriefing of their whole life. Many learn little of the early life of their own parents and then, suddenly, it is too late to ask, unless one has been bold and persuaded them to talk into a tape recorder and reveal their memories.

Companions is a futuristic project, but the Japanese have shown there is a market for anything plausible of this sort, if it reaches an acceptable level of voice and tone and a realistic level of chat. If it does so, then another interesting possibility arises, one close to our original theme of the Internet and life after death. The technical basis of the Companion is a technical matter called machine learning, or what some call data mining: the ability of a computer to learn, within limits, things it did not know before. An obvious successful example is learning to understand or imitate a voice: anyone who buys an IBM typewriter that takes dictation automatically has to train its computer to understand the owner's voice, so as to minimise typing errors.

The same process would allow a computer to imitate its owner's voice: Stephen Hawking's insistence on keeping his twenty year-old electronic voice has had the effect of masking the great advances that have been made. The Companion, after years of debriefing the same owner's life could also, with today's technology, produce a reasonable approximation to their voice. It would also have an organised set of images, emails and documents that tell its owner's life story. So the important question would become: what would to happen to it when the owner died?

Many might prefer to destroy the Companion of a loved one at that point, as some used to put down an aged parent's dogs and others, like ancient Egyptians, killed the wives of rulers at their deaths. But most would not now do that, but would retain the Companion, with its now familiar voice, its memories and detailed knowledge of the loved one, as a powerful and moving memorial, and almost a potted form of the departed.

This is not a very red-blooded life after death on the Internet, but it is something many will find attractive in the future, for themselves, their parents and for their own children. It is no more, perhaps, that a computerised and updated form of the Vidstone or, in a more literary vein, the view of Jules Romains in his prewar novel La Mort de Quelqu'un, that one had some form of existence so long as one was remembered by someone alive and no longer; it is a view not far from much modern rational common-sense.

Ray Kurzweil, the computer pioneer who built the first dictation-typewriter, is said to be devoting his old age entirely to health products so he can stay alive long enough to benefit from what he believes will be the next great technical advance: the reproduction of every human brain cell in a computer, or in silico, as he puts it. The Companion that simulated a dead person, as described above, would be much less radical than that: it might imitate behaviour but would have no tie to any structure in the departed's body or brain. But alas, even Kurtzweil's surviving in silico twin will not be he himself, and will deliver him no more of the traditional afterlife promise than would the survival of his identical twin brother."

Invest in physics -- own a piece of Dirac!

Working...