.
Profile URL | https://www.quora.com/profile/Bill-Paseman |
Question | Answer | Date |
---|---|---|
Which economic system encourages entrepreneurship? | Taking a slightly different tack, an oppressive system that almost crushes the life out of you, make clear that neither you nor your children will ever get ahead, wraps you up in bureaucratic red tape that makes commercial movement impossible, requires bribes to do anything significant, an environment that offers no hope but still offers the chance to migrate to an environment where you have a shot a success. Those environments encourage entrepreneurial-minded people to clear their mind, sharpen their focus and migrate out. I offer as proof the number of first generation Chinese and Indian immigrants in Silicon Valley. Those who grow up in an environment where there is always a chance to succeed lack two important assets that these oppressive environments provide: need and clarity. Many second generation immigrants are quite comfortable thank you, and can pursue entrepreneurship at their leisure, deliberately and intelligently, once they figure out a direction (they certainly don’t want to go the wrong way) and once they have time to follow it (they need to make a living after all). A drowning man is not confused about the direction or speed he needs to swim. A man on the surface can contemplate direction and speed at his leisure. | 1/22/2018 |
Which entrepreneurs, individuals, companies, funds, etc. are actively seeking to fund tech startups in the financial services sector and what is the best way to get in contact with them? | Draper Fintech Connection is one | 1/24/2018 |
Which is more worth it: a CS degree from my state school (UConn) with little to no debt or a CS degree from a school like Purdue or UT Austin with a lot of debt? | See if you can go to the state school and transfer to Purdue your last year. If so, you will need to make yourself attractive to Purdue, so make sure you spend the three years well. | 11/19/2017 |
Which language should I learn at an advanced level for research in computer science like computer vision, digital image processing or machine learning? | Many talks I have seen recently concentrate on how one scales known ml algorithms. Both berkeley's caffe and baidu research spread their algorithms across gpus. As such, I'd probably concentrate on architectures first and languages second. | 1/4/2015 |
Which one are the best startups working in the artificial intelligence field where to do an Internship? | You probably could get a lot more leverage by completing some project on your own. If you don't have any ideas of your own, look at past and current kaggle ( The Home of Data Science) competitions. | 4/19/2016 |
Which school is better: MIT or Caltech? | Joseph Weizenbaum once gave a lecture in MIT's introductory AI class on "Why we will never understand how the brain works". He started with this anecdote: "Suppose you were an electrical engineer who stumbled across a computer in the desert; and further suppose that you know everything there was to know about physics and electronics but knew nothing about computers. Well, that's rather hard to believe..... So suppose you were from Caltech..." That was about 40 years ago, when (Caltech) professors Mead and Conway provided some great new ways for people with no semiconductor experience to do VLSI design. (Which I had the honor of applying to my music synthesizer work). Although Weizenbaum, of course, exaggerated and although it was almost half a century ago, all in all I've found that Caltech in general believes that understanding the fundamentals is critical in any field, and so tend to be more science oriented. That understanding of the fundamentals is what made Mead and Conway's design rules possible. However, there is only so much time in a day, so the temptation is to spend less time on more specific applications. MIT was actually founded by an engineer who tired of dealing with Harvard grads who knew the science but had no application experience. The school's motto is "Head and Hands", which meant that he expected graduates to not only know the science, but how to apply it. However, there is only so much time in a day, so the temptation is to spend less time on more general theory. Of course there are world class theorists at MIT (e.g. Rivest, Shamir Adleman when I was there). But they also were able to apply theory profitably to application (RSA data security). Both are superlative schools. You are lucky if you have a choice. | 4/30/2016 |
Which seminar is best for teenagers? | Best: I don’t know. Most impactful: likely one they pick themselves. Recommendation: Teenagers: get broad experiences. Parents: Help them logistically to try stuff out but make no comment (talk as little as possible unless asked specific questions). Note the uncomfortable truth that a teenager’s peers influence them more than teachers/parents. ←No need to read beyond this point—> Some experiments I tried.
Bottom line, most teenagers I’ve met are not idiots. That said, they are likely to behave self-destructively if a) a parent/teacher pushes them or b) a friend suggests something stupid. When I was at my best, what worked for me was to show them the parts of the world I thought were splendid, then step back, shut up and let them form their own opinions. Hope this helps. | 5/8/2018 |
Which sequence of stack instructions will you use to rotate the content of registers R0, R1 and R2? | you can also do it with just one extra register (a) and 6 Xor statements. Here it is in python (^ is Xor) def bin(s): return str(s) if s<=1 else bin(s>>1) + str(s&1) R1 = int('0110', 2) # TBD: '0101' R2 = int('1110', 2) # TBD: '0110' R3 = int('0101', 2) # TBD: '1110' print 'R1',bin(R1) print 'R2',bin(R2) print 'R3',bin(R3) print "R1 <-> R2" a = R1 ^ R2 R1 = a ^ R1 R2 = a ^ R2 print 'R1',bin(R1) print 'R2',bin(R2) print "R1 <-> R3" a = R1 ^ R3 R1 = a ^ R1 R3 = a ^ R3 print 'R1',bin(R1) print 'R2',bin(R2) print 'R3',bin(R3) produces: R1 110 R2 1110 R3 101 R1 <-> R2 R1 1110 R2 110 R1 <-> R3 R1 101 R2 110 R3 1110 | 11/20/2014 |
Which US based VCs and angels invests seed funding in eSports startups? | Kleiner Perkins did in the 90s | 9/9/2016 |
Which would you choose, a 100,000$ job in Big Oil or a PhD in artificial intelligence at an Ivy League school? | Assuming the oil job is in a lab, try to do both. E.g explore new methods of examining well logs as a thesis topic. School gets the logs. Oil gets the method. You get the phd. | 12/6/2016 |
Why are all major tech companies open sourcing major parts of their AI technology? | Why open source anything? Cost reduction. Here is a more in depth analysis: Deep Learning for Everyone – and (Almost) Free | 1/11/2017 |
Why aren't biotechnology companies hiring more? | In short, for the same reason that relatively few astronauts are hired, despite the quantity of qualified applicants. The barriers to founding a spacecraft company are so great that relatively few such companies exist, and so the number of astronauts is also small. More detail below. ================================================= To your point, my son started UCSD in Bioinformatics, but then changed to CS when he discovered that there were at least 10x more CS jobs advertised at a local Job Fair. Also, when my daughter left Cornell, the best Bay area Job she could scare up paid 2.5x less than a job at Apple. So not just quantity of positions, but also entry salary. Since I was one of the people who guided them into the Biotech field, I was felt pretty lousy. I should have known better. Speaking as an Angel investor, it is simply tougher to come up with a biotech business plan that requires less investment and pays out more on a risk adjusted basis than an IT (e.g. social startup) startup. The old saw is that the best way to make a small fortune in Biotech is to start out with a large fortune. A big part of the reason is that most biotech companies impact human welfare in some way and so require some equivalent to an FDA certification in order to get off the ground. This is not because the FDA is the bad guy. Clinical Trials are expensive and IT startups are less likely to kill someone because of mis-use of their product. This adds to greater marginal expense in established ventures as well. | 2/13/2018 |
Why do mathematicians want a formula that can find prime numbers? What real world problem would that possibly help in solving? | Cryptography, Public-key cryptography in particular. Among other things, Cryptography is used for transmitting money between banks. | 2/28/2015 |
Why do teachers make students do difficult math and science problems by hand, when in real life technology can handle the hard problems and get the answer right in a matter of seconds? | Wow. Diatribe below, but my experience is that those who understood and invent tools (like math) are in charge of their own fate. Those who can only use tools invented by others are at their mercy and often are treated as replaceable parts. =================== When I was 17 in Houston (an oil city), a 17 year old Saudi visited our classroom. He spoke Arabic, French, English (and I believe Italian). He was poised, confident and a little bored with us. I was struggling with German and so I asked him one question. “How many people over there in your school speak that many languages?” His answer “All of them”. In a flash, I realized at that point that going forward, this guy could take my future job. So I buckled down. I went to college, then grad school, then Silicon Valley. They were looking for talent and fortunately, I could deliver. But all the time I was competing with people all over the world who were just like the Saudi, people who had mastered all prior tasks (like math), just as I had. And the ones who advanced were those who mastered every new task given them. It started with school, then delivering economic value to the company (usually engineering, which requires understanding math, not just doing it), then management, then running your own P&L (either a startup or a corporate division), … Perhaps you could view it like a video game, where you need to master challenges at each level before advancing to the next (school, college, first job, …). Challenges at each level are different, and when you advance far enough, you are free. If you screw up too much at the early levels, you won’t be allowed to advance at all. So. Good luck getting your head straight. | 1/14/2018 |
Why does a CS PhD student need to go through programming interviews in a big company? | If I understand correctly, why do an interview at all? Why not just send an offer? Because the company wants to see if there is a fit. Building a company is not like building a brick wall where every piece looks the same. It’s more like building a stone wall, where each stone is unique and ought be compatible on its position to the stones around it. “Fit” is both technical and personal. | 12/6/2017 |
Why does a glass prism absorb UV and infrared radiations? | The fact it is a prism is incidental. Most materials either absorb or transmit radiation (some reflect). As the chart in Glass and solar radiation shows, some glasses transmit Infrared as well. Absorption is a function of the structure of the material's electron orbitals. To be absorbed, there must be quantum energy level pairs that match the photon energy of the radiation. If these energy level pairs are not present, the material is transparent tot he radiation. | 4/28/2016 |
Why does Java use both compiler as well as Interpreter? | Pointers and Garbage Collection - Since not all pointer references can be tracked at compile time, and not all machine architectures have low level support for detecting buffer overflow or bad pointers, you need a runtime check to avoid Buffer overflow hacks in any broadly distributed commercial system. Even ignoring malicious attacks, programmers often inadvertently do poor memory management, so systems like Lisp and Java do automatic memory management (Garbage collection (computer science)). Again, given the need for runtime checks one can either wrap all naked references with compiler generated checks, or avoid the problem entirely by creating an abstraction completely divorced from the underlying architecture. Virtual machine like Parallels and VirtualBox provide this "in the large", JVM "in the small". UNCOL - Expanding on Javed Dosani's comment, since at least the 50's, it has been recognized that compilers are tough to write and update. One solution to this is an UNCOL. To quote the wikipedia article "UNCOL (Universal Computer Oriented Language) was a proposed universal intermediate language for compilers introduced by Melvin E. Conway in 1958.... UNCOL was intended to make compilers economically available for each new instruction set architecture and programming language. Each machine architecture would require just one compiler back end, and each programming language would require one compiler front end." This addressed the issue that for L languages and M machines, one would have to write L*M compilers if one wrote a compiler for each machine. With UNCOL, one write L+M compilers. The ability to port the results of one compiler to many machines is the reason we see other languages now generating code for the JVM (List of JVM languages). However even in the case where there is just one language (Java), you can consider javac (Java compiler) as being the "machine independent" part of the java compiler and jvm as being the machine dependent library part of the compiler. So for every new machine, only the jvm must be re-written. Library Portability - This approach allows allows libraries developed on any system to be used on any other. Note that despite all this rationale, native code java compilers exist. Primarily due to space/speed restrictions on the target devices. | 1/11/2015 |
Why doesn't MIT offer deep learning classes? | or this class? | 5/22/2017 |
Why don't international investors put their production facilities in Nepal? How can investors be attracted? | This is not a new problem. Southern States in the US have worked to attract manufacturing jobs for a long time. Investigate which were most successful and copy their methods. | 5/8/2018 |
Why is calculus considered disturbing and confusing by most students? | Well, I haven’t met anyone who considers it disturbing. It can be difficult. In particular, reading calculus proofs is very different from reading comic books and many of the proofs become “intuitively obvious” only after weeks of hard study. If it is any comfort, two of science’s greatest minds (Newton and Leibnitz) could only create it by building on centuries of prior work. I say science and not mathematics, since it was primarily inspired by mechanics (physics) back when the boundary between science and math was quite fuzzy. (In fact, the combined discipline was called “natural philosophy”.) It was invented to explain physics (mechanics). That is why physics and calculus are such a good fit. One was made to explain the other. As such, many think it best to teach mechanics and calculus together. The canonical example is throwing a Ball up in the air to illustrate integration (going down) and differentiation (going up). Even its inventors can misapply it. It is useful to know that even Newton could screw up the application of calculus to physics. Consider F = dp/dt Where F = force; p = momentum (m*v); and t = time. The correct derivation is: F= m * dv/dt + v * dm/dt The first term is the familiar force = mass times acceleration, since dv/dt is the definition of acceleration. Now, since Newton invented calculus, and in particular invented calculus to explain physics, he probably wrote the above differential at some point. If he did, he probably discarded the second term, assuming mass was a constant and so dm/dt = 0. This mistake by one of the greatest minds on the planet was due to the fact that he didn't know about experiments that lead to the Lorentz transformations (e.g. m = m0/(1 - (v/c)**2)**1/2) Calculus is “rich” The idea of using operators to provide new physical perspectives on problems (moving from position to velocity to acceleration and back again) is “new” and allows new insights. (Similar to moving from polar to cartesian co-ordinates). Personally, this level of symbolic computation was my first taste of the power of formal systems; and so gave me a greater appreciation of the ability of computers to set up, reason about and solve problems which have formal languages. This ability to transition to new worlds may be what some find “disturbing”. Hope this helps. | 1/5/2015 |
Why is distributed computing important for deep learning and machine learning? | Because computing answers takes a long time. Fortunately several MI algorithms (e.g. DL) distribute well and so if the problem takes 12 weeks on one machine, you can find the answer in 3 days with 24 machines. | 12/16/2017 |
Why is it that engineers are pessimistic about automation but economists are optimistic? | I generally agree with Vicki Delaney. One case it might be true is where the engineer sees his job going away and the economist doesn’t. Kind of like NAFTA, great for a manufacturer who can get inputs cheaper. Lousy for the supplier who got replaced. Engineers in both places. The Economist sees general welfare increasing. | 3/27/2018 |
Why isn't there an AI for data cleaning? Doesn't it follow simple logics can be implemented through script? | I believe Paxata does this. | 7/27/2017 |
Why should trigonometry be taught in Secondary School (KS3/4)? How can a teacher make trigonometry an engaging topic for students? | Why not try History? History of trigonometry - Wikipedia Not sure, but as I recall, trig and geometry arose out of very practical considerations. The Nile Valley was fertile and provided an existence for a whole civilization. How to govern it, and more importantly, how do you come up with an objective approach to dividing the land on which the nile flooded? It took a while for -very- early mathematicians to stop concentrating on the sides of triangles (and their ratios) to considering angles (Hipparchus - Wikipedia). And once you had angles, you could consider triangles where the sides could not be measured. E.g. the angle distended by the sun at different times of the day. That geometry helped them create the greatest civilization of their time. An early episode of History channel’s Vikings shows that those sailors were able to keep on constant latitude by taking a cork drawing a circle on it, putting a toothpick in the middle of the circle and floating the cork in a bowl of water on a ship. So long as they kept the toothpick’s shadow in the circle at the right time of day, they could keep the same latitude. According to the TV show, that enabled them to invade England. Due to technology (a cork and a stick) and a simple trig observation. In Master and Commander, Russell Crowe teaches midshipmen to determine their position at sea. “Bring the sun down to the horizon. When its lower limb is touching the horizon... Williamson, look to your sextant! When the orb is no longer rising... then it has reached its zenith and that would be noon.” The sextant allowed England to direct its navy and master the world for a while. See also Figure Out Where You Are With Nothing But a Watch and Protractor Mastering the world required creating spherical trigonometry. Trig on a sphere, where the sum of angles in a triangle are greater then 180 degrees. A practical application of changing Euclid’s parallel postulate. Another application? General relativity, where in 1919 the observers could see the light of a star that was -behind- the sun. -Not- because light bends, but because the geometry of the universe is Riemannian (Spherical), not Euclidian (flat). And then there was the math needed to direct spaceships to the moon, and now, to Mars…. | 5/8/2018 |
Why should you trust people with your business? Workers? | It may not be. Ideally you would have no employees. Full Stop. Instagram had relatively few employees and got acquired for a lot. However it is tough to scale something by yourself. If you can, you should do it. If you can't, you need a structure in place. Partnerships are one alternative to a typical corporate hierarchy. | 5/1/2016 |
Why some companies ship commodities by Rail? | Warren Buffet and Bill Gates have famously invested in rail companies. The reason is that “Each train can move oneton of freight 423 miles on a single gallon of diesel fuel.” | 7/13/2016 |
Why was neural network perceived to be uninteresting before the success of deep learning? | Head to Head Results. Here are the results of the last several Imagenet competitions. Prior to 2012, the best people could come up with was about a 75% correct classification. Hinton was able to close the gap by 10%, which was huge. Since then all approaches have been NN based. | 3/1/2017 |
Why won't Google instant results fill in the search bar when I click them? | Have you tried looking at preference? | 7/13/2016 |
Will first time founders with great team and idea get funding prior to product development? | Absolutely. At some valuation. In fact, knowing nothing further, I’ll give you $0.25 for a non-diluted 50% right now. Being a little less ‘flip’, it is the job of the founder(s) to reduce risk in their ventures. They will get funding along the way. At those funding points, the value is (roughly) proportional to the amount of risk they have reduced. | 5/30/2016 |
Will getting an MBA make me rich? | It can help, but it is usually part of a larger plan. Say you don't like startups. This Business week article may be interesting to you. (MBA Salary Expectations: Sober Reckoning or Wishful Thinking?) It reflects common wisdom (probably correct) that the school you go to matters. However as mentioned elsewhere, the trip up the corporate ladder can take longer than you like, and you have to be pretty good, pretty patient and pretty lucky to get near the top. Say you like startups. MBAs abandon Wall Street for Silicon Valley is a discussion about the Finance and startup trends. I know one Stanford MBA graduate (Vinod Khosla) who founded Sun Microsystems after leveraging the connections he made in Stanford (Andy Bechtolsheim and Berkeley grad Bill Joy). I am primarily technical and so hired corporate management once my (two) companies reached a certain size. Many of the candidates for positions had MBAs in their history and, so long as they had requisite experience, Harvard/Stanford MBAs got my attention. Again, school mattered to me but experience mattered more. And although school and experience were usually correlated, I distinctly remember one Harvard MBA who failed the "test drive". He would not cancel a vacation trip to Paris to attend a customer call with me early on. That "ego trip" cost him about $20M. So, long story short, create a plan that is realistic for you and assess if an MBA is part of it. | 1/7/2015 |
Will Google profit in an Ai first world? | They already are. | 7/26/2016 |
Will the lack of business courses in my degree lower my chances of launching a successful startup? | I helps to acquire the information. Other ways beyond taking the courses is just reading the material (e.g. P&A: Pitch Checklist which I did for my startups) and (free) online courses that you can blow through in a few intense weeks. | 5/2/2016 |
Will there be any Google Science Fair in 2017? | 4/22/2017 | |
Would it be better to get a master's in CS to learn more about AI, combined with security, or simply graduate with a BS and work on gaining that knowledge on my own? | For me, the value in getting a masters (at MIT) was being able to talk to academic minds in person about the subject that interested me (AI), and to network with grad students who had common interests. I.e. the value was in social connection, not in reading particular books or taking particular courses absent personal interaction with the professor. Less of an opportunity to do that in industry. | 2/20/2018 |
Would Theranos have raised yet more investor money had the WSJ not investigated the company? | People, including investors, don’t generally do necessary levels of investigation on their own for key decisions, so I believe that the answer is “yes”. My support is that the WSJ investigation is just a special case of the work typically done by -all- journalists about 15 years ago (Politics vs. cute cats: John Oliver warns about real journalism's demise). | 8/10/2016 |
Would there be a big chance if a startup from Europe went to Silicon Valley to raise money? If yes, how should we go about it? | I suspect you need to supply more information to get an answer. You might look at AngelList to gauge interest. Paseman and Associates: Pitch Checklist lists what I consider an ideal pitch (below). If you can get anywhere close to this presentation, I doubt any VC will care where you are from. Market Risk - Familiarity and Size: I know the Market. It is big. We can get a big piece. Here’s why. Market Risk - Competition: I beat the Competition. I will continue to do so. Here’s why. Product Risk - Development: I have developed the Product. Let me demo it for you. I know what to develop next. Product Risk - Penetration: I am selling the Product. Look at how fast our sales are ramping! Product Risk - "Stickiness": I have repeat business. Look at our repeat order rate! Customers love the product. Call them. Financial Operations Risk: I am Cash Flow Positive. My margins are great. I can be profitable soon. Team Risk: You know, respect and trust the Team. Help me hire more great people. | 10/20/2014 |
Would you be willing to have your brain physically connected to an AI system? Why? | I am already connected to systems like Siri. It is quite convenient. | 12/17/2016 |
Would you continue with your startup when two other startups just launched the same idea as yours? | Well. I would actually need a bit more information to judge. However, One upside is that you may have just validated the market. In addition, if these guys all have come into the market quickly, it may be a sign that the market is growing quickly. I was in a similar situation. The competitor had grown to $35 in sales in two years and I positioned myself as an alternative. We were always smaller, but did just fine. If the market isn't growing quickly, you have to ask if it can support three companies; and either way, you will have to find a differentiating position. For what its worth, although it is unnerving, this will become an ongoing process anyway in any reasonable market (The Illusion of Product/Market Fit for SaaS Companies - Feld Thoughts). Off the cuff, you sound like you just paid for a ticket to ride the horse, if it doesn't cost too much, I'd ride for a while and see if I could stay on. | 1/20/2015 |
Would you give up financial funding of a project of yours if it turned out to be too troublesome? | Hasn’t happened yet. Usually the entrepreneur comes to the conclusion that the project is a dead end and quits. | 11/28/2017 |
Would you like to be on social media by any means of artificial intelligence or something after your death? Why or why not? | No. With apologies to Buffy, Life is for the living. Bradbury had a couple of stories on this. E.g. night call, collect. | 12/28/2017 |