What I absolutely can't understand is the argument that Machine Intelligence needs parallel processing and speed doesn't matter.
If speed is not the problem (no time critical computing) every CPU can be virtualized. It is just a _newthread away.
And the argument, that context-switching is too expensive and at the same time coming up with a platform that has a memory expansion connected over a serial bus (is it serial? at least it's slow) makes me *really* wonder.
If you want a platform that has more memory than Jim's solution, has more CPU-power, draws reasonably power, is at about the same size as 8 Propellers with 16 memory-expansions and costs as much as just two of those memories: Get a mini-ITX board similar to this one: 20..25 Watts including HD. Dual-core-CPU (with an Atom 330) @ 1.6 GHz and a Linux-OS.
<http://resources.mini-box.com/online/MBD-I-D201GLY/intel-d201gly-power-consumption.html>. This will run circles around almost any Propeller-cluster in almost every aspect.
+ cheaper
+ more CPU-power
+ more RAM
+ more permanent storage
+ "unlimited" number of virtual CPUs
- more power (but still not that much at 25 Watts)
And re the argument that context-switching is too expensive:
Just open the task manager in Win (or any other desktop-OS) and see how many complex processes are running.
Now could you explain why the Propeller still is better?
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.: YADRO
I know Dr. Jim wants to use the serial mouse to talk to multiple Prop Proto boards. He has reasons for using multiple boards, for use of additional I/O pins (which you could probably access via your 8 chip system just as well) but also for USB devices for thumb drives, etc.
Mark
He will need host mode to communicate with USB devices. How is he proposing to implement it? It can't be done with a Propeller.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
This whole Dr. Jim thing is something else.· I am a new guy around here but can easily see the history involved with him.· Usually when so many people have such strong feelings against an individual on a forum like this,· there is just cause.·
I tried to watch one of the videos·- why is it a video in the first place?· All I watched was a static screen with some text?· Nothing was said in that video that even closely explained anything at all, however, programmers are not all good speakers.
Is this guy a doctor of something, if so, of what?· If you are going to use Dr. in your title, you darn well better back that up with something!· From the one video I watched, he sure didn't sound like a doctor of any sort.
Perhaps this is the wrong place to voice questions about this person.· Seeing as the magazine (Robot magazine?) is publishing his articles, they should be the ones presented with a lot of these questions.· They are providing him with a platform to talk about a specific science and using his Dr. title (I assume), so I would hope that they have the smarts to verify his credentials.·
So far it seems, based on the comments here, that the magazine has failed in their responsibiilty to the audience.· A publisher allowing someone with questionable credentials speak authoritivly about something is a disservice.
Pehaps he acquired his title in the same way that Col. Sanders of Kentucky Fried Chicken fame obtained his. The eponymous colonel was given his honorary title by the governor of Kentucky. Instead of a secret recipe Dr. Jim seems to have a secret algorithm.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
I think it is enough when we focus on the software, the OS, the hardware.
It really doesn't matter what kind doctor Jim really is. Also not, what postal address he has and things like these. This doesn't make any difference to the products. So please, no more "insights" about that.
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.: YADRO
So now mallred and Dr. Jim don't know that DEC built mainframe computers.· I guess we can add James P. Hogan to the long list of people with significant AI oriented writings they've never heard of.· (FWIW The Two Faces of Tomorrow is considered by many THE novel about creating a Seed AI.)· Hogan got his chops working at DEC in the 1970's.· Selling mainframe computers.· Sheesh.
Also, for all practical purposes a modern PC is a mainframe computer.· Windows NT, which begat Windows 2000, which begat Windows XP, which begat the train wreck called Vista, was itself designed by the people who created DEC VMS.· Which was a mainframe operating system.· All of that bloat in the modern OS is there because of mainframe operating systems.· Mainframes are where we developed techniques for timeslicing, memory protection, and resource sharing that are used on modern desktop PC's.
And calling MS-DOS bloated ... ?· You do realize that your DOS program could simply ignore the operating system, right?· DOS doesn't take up any CPU cycles at all unless you ask it to do something.· I put in·several systems back in the day where I used DOS machines as embedded controllers where all DOS did was boot up the application.· Converting those apps over to Windows was a real nuisance because of the timeslicing.· And nowadays you can put FreeDOS on a modern machine and get the same advantage with gigahertz processors.
If you do not think parallelism is about computational power, you do not know what you are talking about.· While there are costs attached to context switching and the execution of more complex opcodes, those costs are not unreasonable for the convenience of doing your context switching under software control in a single shared memory space, and given a bit of fudge factor for that a fast single machine is exactly as capable as a herd of slower parallel processors.· This equivalence of computational power is a basic tenet of information theory and there is nothing you are going to discover that is going to change that, any more than some genius mathematician is suddenly going to produce a theorem that reveals 2+2=5.· Really, this is very basic stuff that goes all the way back to Alan Turing.
Look, Dr. Jim is NOT LOOKING FOR PERFECT AI! If he is smart at all he knows that it IS NOT POSSIBLE! He is designing a ROUGH artificial intelligence that is not capible at high speeds and does NOT have an answer for EVERY question. The point is that he is wishing to create something that can do AI in the simplest form, and he has choose a Propeller for the task.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-) ·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order. Cheap·shipping unless specified!··········150 left!!··
Here is a list of things we have done, and things we propose to do in the immediate future.
1. KISS Debugger - completed
2. Servo Control software - completed
3. Memory Expansion Kit for Parallax Proto USB board - completed
4. KISS OS - operating system requiring at least one memory board - completed
These items have all been completed. They are currently on the website for sale, except for KISS OS which will appear shortly.
In the works we have:
5. Voice recognition software which will be added to the KISS OS when complete.
6. Machine Intelligence software - already complete, but needs to be ported to our new platform (Propeller chip with KISS OS using SPIN code and Assembly Language, which is what the Propeller understands natively).
7. A new language for the Propeller with only 16 instructions. It will be RISC-like in its simplicity and power.
8. Machine binocular vision.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
What might make many people here less sceptical is a demo of the "Machine Intelligence software" as it exists on what ever platform it runs on. Perhaps in a video. A hint of what it can do would go a long way.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
@microcontrolled -- I know you're trying to give Dr. Jim the benefit of the doubt, but we're having this food fight partly because some of us are becoming convinced that that benefit is misplaced.
As Leon said, Dr. Jim has made some pretty large claims, and has given those of us with some basis for evaluating those claims absolutely zero corroboration. He is demonstrably ignorant of basics of the biological model, which he claims to be the basis for his work. He is unaware of or refuses to acknowledge the work of other Propheads who are doing things like languages and memory boards that parallel his efforts. His justifications for using the Prop instead of one of the many more capable platforms for such work display a basic ignorance of information theory. And he is charging a ridiculous amount of money for his wares, by mallred's own admission "sight unseen" to the buyers, largely on the basis of these extravagant claims.
As for "rough" AI vs. "perfect" AI ... Dr. Jim has made specific claims. I've been working in industry for 25 years and in my experience "rough" and "perfect" aren't real good ways to evaluate a solution; "works" and "doesn't work" are much more descriptive. Dr. Jim has claimed that his creation will learn natural human languages on its own and learn to ask sensible questions. If he gets that in the "works" column he will do a lot more than sell some Propeller expansion boards. And it may be that that is his honest long-term plan. But that's not "rough" AI, that's something that billion dollar companies have launched ocean liners of money at. I am a big fan of the little guy in his garage outsmarting the big money guys myself, but that guy in his garage needs a serious edge, and those of us who have some experience in the field are saying that the Propeller, nifty as it is, is not that edge for this particular problem.
So, localroger, am I to understand that you purchased some of his products, and now are saying that the products do not work? I guess maybe you should submit a short review, so the rest of us, that are not as smart or clever as you are, will know what to do. With your vast "25 years" of knowledge, I bet you have the answers for everything. I believe that most of the people that are viewing this thread are not stupid, so, you can ease up on the protective services.
As for your "biological model" statement, I think that you are "demonstrably ignorant", you want to prove us wrong? Why don't we just stop with all of the innuendo, and stick with the facts that have been presented, or better yet, buy the products, and then you can proclaim "I told you so!".
64 props = 512 independent nodes minus intercommunication nodes, and those could do both things potentially.
So, let's just take some very simple, brute force assessment of compute power. 3 Ghz cpu / 512 threads, leaves ~6Mhz of compute per node compared to the 20 Mhz of compute per propeller node.
Given a totally parallel problem, that fits within the prop constraints, that's only a factor of 5 improvement, and that's ignoring intercommunication with external RAM and between Propellers, and ignoring the kernel required on a sequential computing device.
For a proven solution, that's enough to potentially warrant custom hardware. For development of an unproven solution, it isn't warranted. Could be, but isn't mandatory.
More importantly, that puts the scale of the Propeller hardware on the table here well within the scale others used to explore these kinds of things.
Result = Propeller brings nothing to the table that adds value not obtainable somewhere else. It's cool, but not a value add.
So then, where is the value? I watched the video hoping to answer that question. IMHO, that is the core source of tension here.
Malred: I can put that into business / marketing terms easily enough. When people see a list of differentiators and benefits, sans a value proposition, their spidey sense goes off in the direction of snake oil. When that is refuted with ad_hominem, shot gun style dimunitive responses, you have a recipe for people to be offended, and or at the least seriously confused and frustrated. From a sales perspective, keeping the value problem at issue completely off the table, it's worth noting that confused people do not buy anything. From a product launch / promotion perspective (assuming the snake oil isn't the product, and I'm nearly convinced this is the case, but willing to entertain otherwise), this is absolutely toxic. In this, assuming again that value proposition is not an issue because of IP concerns (and that's a gift friend --a real gift, do consider returning the favor with a genuine expression of the value proposition, if only under strong disclosure terms), you do the Dr. a dis-service in your approach, that warrants some reconsideration. I've been in the product evangelist role multiple times, and I'm putting that on the table out of respect for the quality of the community here, and out of consideration for you and your current role. Translation: Potatohead is being nice and rational about it. Again, do consider returning the favor huh?
To be clear, I'm posting out of morbid interest, and am going back to working on text driver code, with no comments here until said value proposition is at least explained in simple, high level terms.
Have fun Gents!
James P. Hogan is one of my favorite authors. Great characters, interesting science fiction, and deffo a bright light on speculative fiction AI efforts. Mention of his name sparked the desire to comment on this thread.
That's worse [noparse]:([/noparse] -->putting any potential for a value add largely off the table. Prop is at a 2:1 disadvantage in terms of simple compute then. A more real time interaction would be possible with an ordinary Intel CPU + OS / Kernel managing the parallelism.
Edit: I should clarify. It has been stated that compute speed is not a key enabling technology requirement for this "solution". And that's in quotes because I don't see the solution, not to be dimunitive, just rational.
The very simple metric I am reasoning with ignores a lot of things, but does speak to what is possible --what is in scope given the discussion at hand. And the reality is the differences in technology here (prop -vs- traditional sequential compute CPUs), isn't enough to yield a value proposition in and of itself, and that's the issue. "Where is the value add?" It's a simple question, and one that can be answered in high level terms without significantly impacting whatever IP may be on the table. That's all.
Rsadeika said...
So, localroger, am I to understand that you purchased some of his products, and now are saying that the products do not work? I guess maybe you should submit a short review, so the rest of us, that are not as smart or clever as you are, will know what to do. With your vast "25 years" of knowledge, I bet you have the answers for everything. I believe that most of the people that are viewing this thread are not stupid, so, you can ease up on the protective services.
As for your "biological model" statement, I think that you are "demonstrably ignorant", you want to prove us wrong? Why don't we just stop with all of the innuendo, and stick with the facts that have been presented, or better yet, buy the products, and then you can proclaim "I told you so!".
Ray
The RAM expansion hardware simply isn't anything special, and seems very overpriced. They could have gone for surface-mount construction and included a Propeller chip and regulator, and produced something a lot cheaper, smaller, and not requiring any assembly by the user.
Dr Jim's biological model doesn't seem to have been described in detail, and what has been discussed is extremely simplistic. He mentions "wavefront propagation" in isolation at one point in the final clip, with no explanation of why it relevant to his AI technique. Perhaps he is referring to it in connection with a holographic biological mechanism.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Rsadeika -- No I have not purchased any of Dr. Jim's products.· I have examined every thread that has ever been posted here about memory expansions -- and that is a LOT of threads, let me tell you -- and I can tell from that, and Dr. Jim's claims, that whatever he has done (assuming he has done anything other than concoct snake oil) is not all that special.· The Prop offers a fairly straightforward tradeoff between pin usage and bandwidth, and the landscape of that tradeoff has been very vigorously explored by very motivated people.
I don't count my experience in industry as "vast," but it does seem Dr. Jim and I hail from about the same timeframe and some of his claims, such as the importance of mainframe programming experience to his own perspective, are things of which I have direct knowledge.· What I see here is a lot of buzzword dropping, and when they are called on exactly what is meant by those buzzwords the explanations never make any sense.
It's not even like this would be hard.· In the KISS OS thread a number of simple, straightforward questions were asked, like "what does it do."· To give an example of a response that does mean something, here is how I would respond to someone asking about the project I am now working on:
"It uses a SD card for mass storage and provides an interface for multiple cog processes to simultaneously access an arbitrarily large number of open files, optimized for mostly small chunks and a relatively limited bandwidth.· If you want to stream sounds or video you should look at fsrw instead."
Now none of that gives you the source code or a leg-up on trumping me, and I feel safe posting that description even though my boss has decided my work shouldn't be shared with the community.· But Dr. Jim is so coy about his product that similar questions are answered with nothing but vague generalities and promises that Rover will one day get smart enough to ask to shake your hand.· He will not tell us anything substantive about how his product works or what it's supposed to do, other than that it's another step on the road to fulfilling these grandiose claims that have consistently eluded other much better funded researchers, and that's why it costs ten times as much as other competitive solutions.
I too have an ego the size of Jupiter and I've been taken down a few pegs in my time by real experts who stepped in to call me on claims I thought were obvious.· But when I was called I read the references I was given and learned about what went before me, and in some cases I was humbled and in a couple I decided that I was still the big (expletive) dog with the right ideas.· But at no time did I ask people to pay me for things sight unseen on the promise that I could do things that nobody has ever done before, that many people reckon are impossible, and that many more have been trying to do with far more resources than I could ever bring to bear on the problem.
People have been trying to do what Dr. Jim claims he can do roughly since Thomas Ryan published The Adolescence of P-1 in 1977, another prominent AI novel (sometimes compared very negatively to Hogan's 2FoT) from the mainframe era.· None of them have ever achieved natural language recognition and there is no reason to think that any of the hundreds (maybe thousands) trying it out as I type this are very close to that.· If you want us to pony up the ridiculous amount of money being asked for a memory expansion board that is comparable to a number of other much cheaper products, you have to say something a lot more convincing than "trust me."
on edit -- LOL @ Parallax I got a bit excited and used an expletive in this post and the board automatically replaced it with (expletive).· Bravo.
Post Edited (localroger) : 8/1/2009 7:01:31 PM GMT
I do not think it fair to judge by what malread has writer. I think we should give dr. Jim a chance to reply before we criticize more. Maybe the problem is marred has no clue what he is talking about. I would like to see dr jim though stop saying he has the only debugger though.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
@mctrivia -- I think it's clear by now that MR is running interference for DJ. DJ has had plenty of opportunity to step in and clarify the situation. It seems apparent that MR and DJ live either very near or together, and that the MIT not-garage is a space familiar to them both. That DJ would be unaware of this thread is simply not believable. He has chosen not to appear personally for some reason. I can think of some reasons for that, none of which are very flattering, but all of which are quite believable. And most of those reasons suggest that MR is getting hosed.
I agree but there is no point arguing over the words of the guy who says he does not understand. Maybe we can convince the good doctor to keep his promise and answer questions. If not he shows himself twice a lier and we can assume the whole thing a lie also.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
@BradC -- whollly woah, an actual question!?!?!? Yes, IP cameras lag. I was running on the assumption that any practical nascent AI would be running slow. I actually designed an embedded frame grabber (tested to work but never used) to overcome the lack of IP cameras in 2001. If the IP cam latency is too great, there are other pretty low-cost solutions to simply transmit the video to a receiver with a more tightly coupled frame grabber. If you are using Props as the CPU, this gets even easier because you will be throwing most of the information away, and if you can do that before it's sent from the mobile platform to the stationary PC, your bandwidth gets golden.
Phil located 5 of Dr. Gouge's patents. They can be downloaded from freePatentsOnline.com. The most recent one is possibly (?) relevant to the mindset and directlon of the project.
#US 6373984B1 (2002) System and Method for Detecting Patterns or Objects in a Digital Image, James Gouge, Sally Gouge, assigned to Indigo Medical Inc. of Palo Alto CA.
A few random observations:
All but one of the 5 patents are assigned to medical companies in the business of ultrasound imaging, although all claim wider applicability (It is a patent, after all!).
There are 3 external (non-patent) publications referenced, including three articles by James Albus, formerly chief of NIST, A Model of the Brain for Robot Control ... parts I, II and III.
Patents have a background section that describes prior art, in order to set apart the new invention. Neural Networks in this context are denigrated as follows:
"... these systems require extensive training through the presentation of numerous images to the neural network. Because the images are presented to the system without intelligent insight as to the content of the images, the system tends to only learn statistically significant events and ignores image events less frequently encountered..."
and it goes on to give an example about identification of "abnormal cell structures indicative of some cancer".
Mallred's statement about the new approach not being a neural net makes more sense if you read it in the narrow context of patent language. However, the claimed invention would in common parlance be a neural net. It's goal is to analyze patterns in an area of pixels, and what sets it apart is "intelligent insight". :
"The system includes synaptic link generator, a knowledge element (knixel) identifier, a theme identifier, and a corresponding theme/knixel associations knowledge base."
Knixel? That's catchy. The text goes on to describe the elements and how the training is accomplished. I admit that I have not dug into it in detail. It is not my field, and reading patents is thick going, to read between the lines. But skimming the text and in a gestalt of the diagrams, it looks substantive. Far more so that the topic videos of this thread, which I found quite the opposite.
I am left with the feeling that Dr. Gouge may have a rather specific goal in mind, maybe trying actually to reduce to practice the ideas in these patents. I am as mystified as anybody else about how how it is all being rolled out.
Patent 5267328 (1993) is also cited and has an interesting observation about how processing of back-scatter data from ultrasound is typically difficult mathematics. It involves solutions to an inverse problem, that is, working backwards from a pattern to the object and sound sources that produced it. But with the subject invention:
"... these algorithms are not based upon strict mathematical concepts but are rather translated to the computer in terms of bit strings which require very little processing time. The disclosed methods are designed for efficient image processing in real time on PC-type computer equipment".
Again, I don't know if the idea worked or didn't work, but that in itself in context of a a narrow application would be quite an accomplishment. Most of the talk I have heard about imaging does in fact depend on supercomputers solving the inverse problem. I can hope Dr. Gouge's insights into imaging will end up helping anyone who lands unhappily in a hospital ER with a mysterious illness.
Edit: added following references:
Here is a link to an article written by Dr. Gouge about neuroscience as a tool for image processing.
Here are the articles referenced in the recent Gouge patent. Is archival material from Byte online anywhere?
-- Albus, J.S., "A Model of the Brain for Robot Control Part 1: Defining Notation," Byte Magazine, pgs. 10-32, June 1979.
-- Albus, J.S., "A Model of the Brain for Robot Control Part 2: A Neurological Model," Byte Magazine, pgs. 54-95, July 1979.
-- Albus, J.S., "A Model for the Brain for Robot Control Part 3: A Comparison of the Brain and our Model" pages 66-78, Aug. 1979
My question is, if this is so ground breaking, why not be courting the military or big industry, (pick one), and keep it to youself untill it would be ready to reap big rewards? He's not going to make a lot of money off the hobby market.
Even though this thread hasn't generated the the amounts of replies to be considered as a record, the amount of typed words may be. [noparse]:)[/noparse]
Laying Low on this one, but only to hope Mark's wife is doing better.
mallred> Our technology promises to provide a major step in the direction of autonomous, intelligent machines and provides an
> excellent, low-cost platform for future research. Isn't that what it's all about?
> .... As an example, there is one question to answer. Where is the truely intelligent robot? It has not yet been made.
[noparse][[/noparse]As if they are about to.]
Hover1 (Jim)> My question is, if this is so ground breaking, why not be courting the military or big industry, (pick one),
> and keep it to youself untill it would be ready to reap big rewards? He's not going to make a lot of money off the hobby market.
Well, with what Leon and with Tracy Allen just posted, we have this on face value:
1. Mr. Allred is in charge.
2. The patents claim something substantial enough to garner venture capital interest.
Therefore, WHY on earth is Mr. Allred spending so much time here? We're good people, but not (I assume) venture capitalists who could bulk up his organization to move it forward. (And to Phils' earlier point of expansion leading to extra costs, taxes, benefits - that's not true --- they could contract their workers via a company like Paychex.)
A good venture capitalist wants to know what is really under the hood. If it's sound, there will be enough funding to cover such startup and operational costs, etc. The "we're just a one man show" and "it's top secret" (so to say) usually means: "we don't have a sound enough business plan to make it work, because the idea itself isn't viable."
And then consider Potatohead's earlier, LUCID post about the impact Mr. Allred's style of posting, his "marketing presentation," has on the public and on potential investors. It's "Toxic".
As a former technical consultant to venture capitalists, and an investor myself, I would have to say, even *with* a solid, realistic business plan, the marketing effort will fail because the President himself does not present his venture properly. Even if the underlying assets are not toxic, the presentation is.
Something smells more than fishy here.
My virtual $2.00
- Howard
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Post Edited (CounterRotatingProps) : 8/1/2009 9:24:54 PM GMT
I could be wrong, but I think the word is liar, not familiar with lier. I guess I may have been to subtle in my other posts, if you are going to call the guy a liar, then "put up, or shut up"! And if that is to subtle, then how about "show me the proof". And just because it is your opinion, it does not constitute a fact.
So, where are we at now? As far as I can tell we have had a lot of posturing, but nobody has shown me any real proof, there own personal theories, and beliefs, but no proof. I am not trying to take sides, and I am not trying to be a mediator, I like to deal with facts. I guess all the skeptics have one direction left, semantics. That should be a lot of fun.
I am still not sure as to whether I will be purchasing the products, my project is still in the theory state. When I get to the POC then I will re-evaluate the products, and then I will see where I am at. As for the pricing, if it fits into my plans, then I do not consider it to be to expensive. I can see that there a lot of people trying to sell there wares, so I can see why some people would be upset with the competition. But, that is life, live with it.
( I personally am not calling the man a fibber or a Greek Musical instrument - but his tune sure rings hollow in the ears of many here! [noparse]:)[/noparse]
No one here selling anything has real problems with each other's products or competition - we are all basically in the same boat.
In fact, from what I've witnessed, it seems folks here are about as *anti* competitive as can be. Mutual support is the motto.
Perhaps one of the main reasons why this is all so irksome, that and, the lack of substance. ·By way of close analogy: Would you buy an expensive new chip, much less plan an entire production run with it, without ever seeing the chip's datasheet??
Being sold a "Bill of Goods" means you're buying the bill *paper* - not the goods they supposedly represent!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Post Edited (CounterRotatingProps) : 8/1/2009 9:34:09 PM GMT
I don't see any overt deception at all in Machine IntelliTech's public statements, because I'm certain they believe their own hype. My fear, though, is that they may be deceiving themselves and, in the process, setting themselves up for failure.* And that would be too bad, since Mark and Jim both seem like nice people who are sincerely trying to make a mark in the industry. But my suspicion is that they don't yet have anything of substance to show — maybe because it's turning out to be a lot harder than they thought it would be — and that's why all the hand-waving and gratuitous puffery. It would have been better, I suppose, if they'd just kept their own counsel until they had something of substance to demonstrate; but they've chosen the far more dangerous path of pre-emptive bluster. And by doing so, they've set themselves up for the questioning, if not outright derision, extant in these forums. Neither their swagger nor our incredulity will likely subside any time soon. In the meantime, it all makes for good theater, but even that is beginning to wear thin. If they could just toss us a real bone to chew on, rather than more feints, we'd be a lot happier; and they'd garner some respect for their efforts.
-Phil
*Who among us hasn't been guilty of this at least once? I thought for sure that I could win the Netflix Prize, no problem. I mean, how hard could it be to better Netflix's own predictions by a mere ten percent? Never mind that there were hundreds of others just like me — and a lot smarter — thinking the same thing. I even bought a fast PC, just so I could crunch data 24/7. I never came close. Will I ever make the same mistake again? Gosh, I hope so! Life would be pretty dull without an inflated sense of self-confidence!
Yeah Phil, but you werent' selling people shares in your eventual success on the Netflix thing. That's where it goes over the line from fooling yourself to being a scam.
Comments
If speed is not the problem (no time critical computing) every CPU can be virtualized. It is just a _newthread away.
And the argument, that context-switching is too expensive and at the same time coming up with a platform that has a memory expansion connected over a serial bus (is it serial? at least it's slow) makes me *really* wonder.
If you want a platform that has more memory than Jim's solution, has more CPU-power, draws reasonably power, is at about the same size as 8 Propellers with 16 memory-expansions and costs as much as just two of those memories: Get a mini-ITX board similar to this one: 20..25 Watts including HD. Dual-core-CPU (with an Atom 330) @ 1.6 GHz and a Linux-OS.
<http://resources.mini-box.com/online/MBD-I-D201GLY/intel-d201gly-power-consumption.html>. This will run circles around almost any Propeller-cluster in almost every aspect.
+ cheaper
+ more CPU-power
+ more RAM
+ more permanent storage
+ "unlimited" number of virtual CPUs
- more power (but still not that much at 25 Watts)
And re the argument that context-switching is too expensive:
Just open the task manager in Win (or any other desktop-OS) and see how many complex processes are running.
Now could you explain why the Propeller still is better?
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
He will need host mode to communicate with USB devices. How is he proposing to implement it? It can't be done with a Propeller.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/1/2009 10:26:19 AM GMT
I tried to watch one of the videos·- why is it a video in the first place?· All I watched was a static screen with some text?· Nothing was said in that video that even closely explained anything at all, however, programmers are not all good speakers.
Is this guy a doctor of something, if so, of what?· If you are going to use Dr. in your title, you darn well better back that up with something!· From the one video I watched, he sure didn't sound like a doctor of any sort.
Perhaps this is the wrong place to voice questions about this person.· Seeing as the magazine (Robot magazine?) is publishing his articles, they should be the ones presented with a lot of these questions.· They are providing him with a platform to talk about a specific science and using his Dr. title (I assume), so I would hope that they have the smarts to verify his credentials.·
So far it seems, based on the comments here, that the magazine has failed in their responsibiilty to the audience.· A publisher allowing someone with questionable credentials speak authoritivly about something is a disservice.
End of rant.
Chris
·
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/1/2009 1:05:16 PM GMT
It really doesn't matter what kind doctor Jim really is. Also not, what postal address he has and things like these. This doesn't make any difference to the products. So please, no more "insights" about that.
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
Also, for all practical purposes a modern PC is a mainframe computer.· Windows NT, which begat Windows 2000, which begat Windows XP, which begat the train wreck called Vista, was itself designed by the people who created DEC VMS.· Which was a mainframe operating system.· All of that bloat in the modern OS is there because of mainframe operating systems.· Mainframes are where we developed techniques for timeslicing, memory protection, and resource sharing that are used on modern desktop PC's.
And calling MS-DOS bloated ... ?· You do realize that your DOS program could simply ignore the operating system, right?· DOS doesn't take up any CPU cycles at all unless you ask it to do something.· I put in·several systems back in the day where I used DOS machines as embedded controllers where all DOS did was boot up the application.· Converting those apps over to Windows was a real nuisance because of the timeslicing.· And nowadays you can put FreeDOS on a modern machine and get the same advantage with gigahertz processors.
If you do not think parallelism is about computational power, you do not know what you are talking about.· While there are costs attached to context switching and the execution of more complex opcodes, those costs are not unreasonable for the convenience of doing your context switching under software control in a single shared memory space, and given a bit of fudge factor for that a fast single machine is exactly as capable as a herd of slower parallel processors.· This equivalence of computational power is a basic tenet of information theory and there is nothing you are going to discover that is going to change that, any more than some genius mathematician is suddenly going to produce a theorem that reveals 2+2=5.· Really, this is very basic stuff that goes all the way back to Alan Turing.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-)
·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order.
Cheap·shipping unless specified!··········150 left!!··
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
As Leon said, Dr. Jim has made some pretty large claims, and has given those of us with some basis for evaluating those claims absolutely zero corroboration. He is demonstrably ignorant of basics of the biological model, which he claims to be the basis for his work. He is unaware of or refuses to acknowledge the work of other Propheads who are doing things like languages and memory boards that parallel his efforts. His justifications for using the Prop instead of one of the many more capable platforms for such work display a basic ignorance of information theory. And he is charging a ridiculous amount of money for his wares, by mallred's own admission "sight unseen" to the buyers, largely on the basis of these extravagant claims.
As for "rough" AI vs. "perfect" AI ... Dr. Jim has made specific claims. I've been working in industry for 25 years and in my experience "rough" and "perfect" aren't real good ways to evaluate a solution; "works" and "doesn't work" are much more descriptive. Dr. Jim has claimed that his creation will learn natural human languages on its own and learn to ask sensible questions. If he gets that in the "works" column he will do a lot more than sell some Propeller expansion boards. And it may be that that is his honest long-term plan. But that's not "rough" AI, that's something that billion dollar companies have launched ocean liners of money at. I am a big fan of the little guy in his garage outsmarting the big money guys myself, but that guy in his garage needs a serious edge, and those of us who have some experience in the field are saying that the Propeller, nifty as it is, is not that edge for this particular problem.
As for your "biological model" statement, I think that you are "demonstrably ignorant", you want to prove us wrong? Why don't we just stop with all of the innuendo, and stick with the facts that have been presented, or better yet, buy the products, and then you can proclaim "I told you so!".
Ray
So, let's just take some very simple, brute force assessment of compute power. 3 Ghz cpu / 512 threads, leaves ~6Mhz of compute per node compared to the 20 Mhz of compute per propeller node.
Given a totally parallel problem, that fits within the prop constraints, that's only a factor of 5 improvement, and that's ignoring intercommunication with external RAM and between Propellers, and ignoring the kernel required on a sequential computing device.
For a proven solution, that's enough to potentially warrant custom hardware. For development of an unproven solution, it isn't warranted. Could be, but isn't mandatory.
More importantly, that puts the scale of the Propeller hardware on the table here well within the scale others used to explore these kinds of things.
Result = Propeller brings nothing to the table that adds value not obtainable somewhere else. It's cool, but not a value add.
So then, where is the value? I watched the video hoping to answer that question. IMHO, that is the core source of tension here.
Malred: I can put that into business / marketing terms easily enough. When people see a list of differentiators and benefits, sans a value proposition, their spidey sense goes off in the direction of snake oil. When that is refuted with ad_hominem, shot gun style dimunitive responses, you have a recipe for people to be offended, and or at the least seriously confused and frustrated. From a sales perspective, keeping the value problem at issue completely off the table, it's worth noting that confused people do not buy anything. From a product launch / promotion perspective (assuming the snake oil isn't the product, and I'm nearly convinced this is the case, but willing to entertain otherwise), this is absolutely toxic. In this, assuming again that value proposition is not an issue because of IP concerns (and that's a gift friend --a real gift, do consider returning the favor with a genuine expression of the value proposition, if only under strong disclosure terms), you do the Dr. a dis-service in your approach, that warrants some reconsideration. I've been in the product evangelist role multiple times, and I'm putting that on the table out of respect for the quality of the community here, and out of consideration for you and your current role. Translation: Potatohead is being nice and rational about it. Again, do consider returning the favor huh?
To be clear, I'm posting out of morbid interest, and am going back to working on text driver code, with no comments here until said value proposition is at least explained in simple, high level terms.
Have fun Gents!
James P. Hogan is one of my favorite authors. Great characters, interesting science fiction, and deffo a bright light on speculative fiction AI efforts. Mention of his name sparked the desire to comment on this thread.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
Chat in real time with other Propellerheads on IRC #propeller @ freenode.net
Safety Tip: Life is as good as YOU think it is!
No, 64 COGS = 8 Props. That's what mallred said.
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
Edit: I should clarify. It has been stated that compute speed is not a key enabling technology requirement for this "solution". And that's in quotes because I don't see the solution, not to be dimunitive, just rational.
The very simple metric I am reasoning with ignores a lot of things, but does speak to what is possible --what is in scope given the discussion at hand. And the reality is the differences in technology here (prop -vs- traditional sequential compute CPUs), isn't enough to yield a value proposition in and of itself, and that's the issue. "Where is the value add?" It's a simple question, and one that can be answered in high level terms without significantly impacting whatever IP may be on the table. That's all.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
Chat in real time with other Propellerheads on IRC #propeller @ freenode.net
Safety Tip: Life is as good as YOU think it is!
Post Edited (potatohead) : 8/1/2009 5:08:18 PM GMT
The RAM expansion hardware simply isn't anything special, and seems very overpriced. They could have gone for surface-mount construction and included a Propeller chip and regulator, and produced something a lot cheaper, smaller, and not requiring any assembly by the user.
Dr Jim's biological model doesn't seem to have been described in detail, and what has been discussed is extremely simplistic. He mentions "wavefront propagation" in isolation at one point in the final clip, with no explanation of why it relevant to his AI technique. Perhaps he is referring to it in connection with a holographic biological mechanism.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/1/2009 6:26:08 PM GMT
I don't count my experience in industry as "vast," but it does seem Dr. Jim and I hail from about the same timeframe and some of his claims, such as the importance of mainframe programming experience to his own perspective, are things of which I have direct knowledge.· What I see here is a lot of buzzword dropping, and when they are called on exactly what is meant by those buzzwords the explanations never make any sense.
It's not even like this would be hard.· In the KISS OS thread a number of simple, straightforward questions were asked, like "what does it do."· To give an example of a response that does mean something, here is how I would respond to someone asking about the project I am now working on:
"It uses a SD card for mass storage and provides an interface for multiple cog processes to simultaneously access an arbitrarily large number of open files, optimized for mostly small chunks and a relatively limited bandwidth.· If you want to stream sounds or video you should look at fsrw instead."
Now none of that gives you the source code or a leg-up on trumping me, and I feel safe posting that description even though my boss has decided my work shouldn't be shared with the community.· But Dr. Jim is so coy about his product that similar questions are answered with nothing but vague generalities and promises that Rover will one day get smart enough to ask to shake your hand.· He will not tell us anything substantive about how his product works or what it's supposed to do, other than that it's another step on the road to fulfilling these grandiose claims that have consistently eluded other much better funded researchers, and that's why it costs ten times as much as other competitive solutions.
I too have an ego the size of Jupiter and I've been taken down a few pegs in my time by real experts who stepped in to call me on claims I thought were obvious.· But when I was called I read the references I was given and learned about what went before me, and in some cases I was humbled and in a couple I decided that I was still the big (expletive) dog with the right ideas.· But at no time did I ask people to pay me for things sight unseen on the promise that I could do things that nobody has ever done before, that many people reckon are impossible, and that many more have been trying to do with far more resources than I could ever bring to bear on the problem.
People have been trying to do what Dr. Jim claims he can do roughly since Thomas Ryan published The Adolescence of P-1 in 1977, another prominent AI novel (sometimes compared very negatively to Hogan's 2FoT) from the mainframe era.· None of them have ever achieved natural language recognition and there is no reason to think that any of the hundreds (maybe thousands) trying it out as I type this are very close to that.· If you want us to pony up the ridiculous amount of money being asked for a memory expansion board that is comparable to a number of other much cheaper products, you have to say something a lot more convincing than "trust me."
on edit -- LOL @ Parallax I got a bit excited and used an expletive in this post and the board automatically replaced it with (expletive).· Bravo.
Post Edited (localroger) : 8/1/2009 7:01:31 PM GMT
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
Patti Allred, President
patti@machineinteltech.com
Mark Allred, Vice President and COO
mark@machineinteltech.com
Dr. James O. Gouge, Chief Engineer
jim@machineinteltech.com
It looks like Dr. Jim is employed by Mr. and Mrs. Allred.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
#US 6373984B1 (2002) System and Method for Detecting Patterns or Objects in a Digital Image, James Gouge, Sally Gouge, assigned to Indigo Medical Inc. of Palo Alto CA.
A few random observations:
All but one of the 5 patents are assigned to medical companies in the business of ultrasound imaging, although all claim wider applicability (It is a patent, after all!).
There are 3 external (non-patent) publications referenced, including three articles by James Albus, formerly chief of NIST, A Model of the Brain for Robot Control ... parts I, II and III.
Patents have a background section that describes prior art, in order to set apart the new invention. Neural Networks in this context are denigrated as follows:
"... these systems require extensive training through the presentation of numerous images to the neural network. Because the images are presented to the system without intelligent insight as to the content of the images, the system tends to only learn statistically significant events and ignores image events less frequently encountered..."
and it goes on to give an example about identification of "abnormal cell structures indicative of some cancer".
Mallred's statement about the new approach not being a neural net makes more sense if you read it in the narrow context of patent language. However, the claimed invention would in common parlance be a neural net. It's goal is to analyze patterns in an area of pixels, and what sets it apart is "intelligent insight". :
"The system includes synaptic link generator, a knowledge element (knixel) identifier, a theme identifier, and a corresponding theme/knixel associations knowledge base."
Knixel? That's catchy. The text goes on to describe the elements and how the training is accomplished. I admit that I have not dug into it in detail. It is not my field, and reading patents is thick going, to read between the lines. But skimming the text and in a gestalt of the diagrams, it looks substantive. Far more so that the topic videos of this thread, which I found quite the opposite.
I am left with the feeling that Dr. Gouge may have a rather specific goal in mind, maybe trying actually to reduce to practice the ideas in these patents. I am as mystified as anybody else about how how it is all being rolled out.
Patent 5267328 (1993) is also cited and has an interesting observation about how processing of back-scatter data from ultrasound is typically difficult mathematics. It involves solutions to an inverse problem, that is, working backwards from a pattern to the object and sound sources that produced it. But with the subject invention:
"... these algorithms are not based upon strict mathematical concepts but are rather translated to the computer in terms of bit strings which require very little processing time. The disclosed methods are designed for efficient image processing in real time on PC-type computer equipment".
Again, I don't know if the idea worked or didn't work, but that in itself in context of a a narrow application would be quite an accomplishment. Most of the talk I have heard about imaging does in fact depend on supercomputers solving the inverse problem. I can hope Dr. Gouge's insights into imaging will end up helping anyone who lands unhappily in a hospital ER with a mysterious illness.
Edit: added following references:
Here is a link to an article written by Dr. Gouge about neuroscience as a tool for image processing.
Here are the articles referenced in the recent Gouge patent. Is archival material from Byte online anywhere?
-- Albus, J.S., "A Model of the Brain for Robot Control Part 1: Defining Notation," Byte Magazine, pgs. 10-32, June 1979.
-- Albus, J.S., "A Model of the Brain for Robot Control Part 2: A Neurological Model," Byte Magazine, pgs. 54-95, July 1979.
-- Albus, J.S., "A Model for the Brain for Robot Control Part 3: A Comparison of the Brain and our Model" pages 66-78, Aug. 1979
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com
Post Edited (Tracy Allen) : 8/1/2009 10:02:31 PM GMT
Even though this thread hasn't generated the the amounts of replies to be considered as a record, the amount of typed words may be. [noparse]:)[/noparse]
Laying Low on this one, but only to hope Mark's wife is doing better.
Jim
Post Edited (hover1) : 8/1/2009 8:31:31 PM GMT
> excellent, low-cost platform for future research. Isn't that what it's all about?
> .... As an example, there is one question to answer. Where is the truely intelligent robot? It has not yet been made.
[noparse][[/noparse]As if they are about to.]
Hover1 (Jim)> My question is, if this is so ground breaking, why not be courting the military or big industry, (pick one),
> and keep it to youself untill it would be ready to reap big rewards? He's not going to make a lot of money off the hobby market.
Well, with what Leon and with Tracy Allen just posted, we have this on face value:
1. Mr. Allred is in charge.
2. The patents claim something substantial enough to garner venture capital interest.
Therefore, WHY on earth is Mr. Allred spending so much time here? We're good people, but not (I assume) venture capitalists who could bulk up his organization to move it forward. (And to Phils' earlier point of expansion leading to extra costs, taxes, benefits - that's not true --- they could contract their workers via a company like Paychex.)
A good venture capitalist wants to know what is really under the hood. If it's sound, there will be enough funding to cover such startup and operational costs, etc. The "we're just a one man show" and "it's top secret" (so to say) usually means: "we don't have a sound enough business plan to make it work, because the idea itself isn't viable."
And then consider Potatohead's earlier, LUCID post about the impact Mr. Allred's style of posting, his "marketing presentation," has on the public and on potential investors. It's "Toxic".
As a former technical consultant to venture capitalists, and an investor myself, I would have to say, even *with* a solid, realistic business plan, the marketing effort will fail because the President himself does not present his venture properly. Even if the underlying assets are not toxic, the presentation is.
Something smells more than fishy here.
My virtual $2.00
- Howard
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Post Edited (CounterRotatingProps) : 8/1/2009 9:24:54 PM GMT
So, where are we at now? As far as I can tell we have had a lot of posturing, but nobody has shown me any real proof, there own personal theories, and beliefs, but no proof. I am not trying to take sides, and I am not trying to be a mediator, I like to deal with facts. I guess all the skeptics have one direction left, semantics. That should be a lot of fun.
I am still not sure as to whether I will be purchasing the products, my project is still in the theory state. When I get to the POC then I will re-evaluate the products, and then I will see where I am at. As for the pricing, if it fits into my plans, then I do not consider it to be to expensive. I can see that there a lot of people trying to sell there wares, so I can see why some people would be upset with the competition. But, that is life, live with it.
Ray
where's the beef, as they say?·
( I personally am not calling the man a fibber or a Greek Musical instrument - but his tune sure rings hollow in the ears of many here! [noparse]:)[/noparse]
No one here selling anything has real problems with each other's products or competition - we are all basically in the same boat.
In fact, from what I've witnessed, it seems folks here are about as *anti* competitive as can be. Mutual support is the motto.
Perhaps one of the main reasons why this is all so irksome, that and, the lack of substance. ·By way of close analogy: Would you buy an expensive new chip, much less plan an entire production run with it, without ever seeing the chip's datasheet??
Being sold a "Bill of Goods" means you're buying the bill *paper* - not the goods they supposedly represent!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Post Edited (CounterRotatingProps) : 8/1/2009 9:34:09 PM GMT
-Phil
*Who among us hasn't been guilty of this at least once? I thought for sure that I could win the Netflix Prize, no problem. I mean, how hard could it be to better Netflix's own predictions by a mere ten percent? Never mind that there were hundreds of others just like me — and a lot smarter — thinking the same thing. I even bought a fast PC, just so I could crunch data 24/7. I never came close. Will I ever make the same mistake again? Gosh, I hope so! Life would be pretty dull without an inflated sense of self-confidence!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔