Man finds a black kind of rock that burns; discovers that you can get a lot of this rock if you dig deeper, but deep mines have water. In order to successfully mine this rock, man devises a steam powered engine (neatly enough powered by this same rock) to pump out the water. No, not the steam engines you're familiar with. This is the Newcomen Steam Engine: http://en.wikipedia.org/wiki/Newcomen_engine
The Newcomen Engine has a fatal flaw: it cools the steam for the return stroke, losing energy to the latent heat of evaporation each time. James Watt discovers the latent heat of evaporation, and realizes that separating the condenser from the piston would improve efficiency. So let's go build some railroads, right? Not so fast. It would still be another 30 years (100 years from the invention of the Newcomen Engine) before railroads and ferry boats would be regularly powered by reciprocating steam engines.
What's the moral? For 100 years, vast leaps in technology came one after the other. In the process, the Laws of Thermodynamics were discovered and described. Many learned men stood around patting each other on the back at how successful, how inventive they were...at digging a black rock called "coal" from the ground.
But most people don't dig rock from the ground. Most people do travel from point A to point B on a fairly regular basis. The world changed when 100 years of technology left the mine shaft and the factory, and got people where they were going just a bit faster.
I'm convinced that computers are still at the Newcomen/Watt transition. We have a ways to go before the world truly changes.
These ideas, and the idea that they are frighteningly ambitious, clearly come from the personal experience of living in northern California and dealing with tech startups most of your time.
These are largely first world problems. Here are some ambitious ideas:
- distributed power generation that's cheap enough and renewable enough so people in rural parts of sub-Saharan Africa don't have brown outs anymore.
- synthetic food generation a la star trek
- desalination that is cheap enough for a farmer in Mozambique to do himself
There are more, lots more. People outside the valley bubble have real problems.
Solving first world problems gets you first world ROI. Y-Combinator is an investment firm. I'm not judging, I don't have a dog in the fight. But I'm pretty sure that's why you don't see stuff like stopping brown outs in rural sub-Saharan Africa on this list. There's no first-world money in it.
I think the best ideas improves first world issues while fixing third world problems. Energy generation for example, if you managed to figure out how to generate power/propulsion using a cheap and small physical object and a renewable resource you both solves first world issues and third world problems.
Education needs to catch up with the world in the first world, but if you do that in an async manner online then you can at the same time solve the problem with education in africa.
distributed power generation that's cheap enough and renewable enough so people in rural parts of sub-Saharan Africa don't have brown outs anymore.
desalination that is cheap enough for a farmer in Mozambique to do himself
I am sorry, but you can not fix broken states with technology. Want to help Mozambique and sub-Saharan people? Find a way to turn their governments into well working ones. This is at once infinitely more difficult and also the only thing which will really solve the common developing world problems.
The University problem is potentially a "3rd world" problem, especially inasmuch as you interpret it as a learning/teaching problem.
Non consumption is traditionally a good place to start and I would suggest that high school & junior university level is the most disruptive place to start. Non-consumption of senior high school to junior university level education is something in great abundance in poor countries.
There is also probably a more incentive for potential students to play along in poor countries. A 9th-10th year dropout is more likely to be in that position because of access or soem other problem that technological innovation is good at dealing with. More importantly, the ROI on those 2-4 years of education is probably much higher.
If you were to go after job skills/ training as the point of attack (as opposed to general education) poor countries are also a great place to start. For a lot of skills there is demand at the bottom: bookkeeping, graphic design, programming, etc. Bringing a person making <$1-$2 per hour to a point where they can command a $3+ is fundamentally doable. That's a big incentive.
What if we made people in subsaharan africa as rich as europeans & americans? Then the solution to "brown outs" there is the solutions to brown outs here. Better infastructure.
The problem with search is that not only is Google getting worse, but I've also mostly outgrown it, in that it isn't sophisticated to answer pretty much any scientific question I would want to ask.
- No way to search for a scientific question and get a summary of the current scientific consensus or viewpoints on specific issues
- It's really hard to access academic journal articles online.
- Even when you can access journal articles, it's hard to know which ones to look in to answer your question. Sometimes it's hard to even know which field(s) your question falls under.
- Even if you vaguely know which field your question falls under, you don't necessarily know any of the vocabulary used by that field.
- No way to search by dependent and independent variables, confounding variables, etc.
- No way to sort articles by the quality of their methodology, the quality of the journal they were published in, the quality of the researchers, etc.
I know this isn't a product that more than 1% of the population would use, but if someone built it then maybe there are other things it could be used for.
You're talking about highly-focused, or micro, search. Yeah, Google doesn't seem to do that very well. They have a few segments, like book search and image search, but it's not specific enough.
One thing I search for sometimes are code examples in a particular language. Search for something in C on Google and you end up with lots of stuff for C++, C#, etc... Github, with its large repository of public code, lets you filter by programming language, and is much better than Google in these cases.
Bing copies Google. DuckDuckGo returns different things than Google, but otherwise is a copy. There's no micro search engine for specific topics and sub-topics, outside of site-specific search. Market opportunity...
Does it bother anyone that "frightenly ambitious" begins with search and email? Seriously? This is the pinnacle of our contribution to mankind - building search engines and to-do lists?
Where's the lunar base? The flying car? The personal robot? The cyborg? Meh, maybe I'm just getting old and grumpy. (To be fair I did like the other ones).
Search is another word for artificial intelligence.
Email means a universal way for people to communicate.
Lunar bases & flying cars aren't ambitious, they are just expensive.
Personal robots are here now in vertical spaces[1] and there are at least 200,000 cyborgs walking around now[2]. I think both these areas are worth working in, but I think you underestimate the ambition of a word like search.
That was just a selection of ambitious ideas. I didn't mean to imply it was a complete list; I never imagined anyone would think I meant that.
I almost added a section on robots, but the talk was already long enough and I figured 7 was enough to illustrate my point that ambitious startup ideas are frightening.
Facebook and Google have improved the daily lives of billions of people. Tesla Motors' electric car? Very impressive and a lot of promise, but so far all it's done is increase the self-satisfaction of ~1000 people.
Look at Star Trek. Other than the spaceship itself the most impressive things are the technology the people use. The Siri-like computer interface, iPad-like computers, Skype-like communications technology, etc.
The frightening part is the idea that you would be trying to disrupt seemingly invincible tech companies like google, which would be an insane business plan for a startup looking for a $50k kickstart investment.
But, I remember when google was first released. I thought the search engine market was already full with yahoo and alta vista dominating at the time. Who knew?!
you are vastly underestimating how much email has changed the world. if a better email can do for this generation what email did for the last one, i'll back it against anything on your list for sheer bang for the buck.
The lunar base and the flying cars are pointless if your goal is to improve people's lives instead of just satisfying egos.
And the personal robot is already being worked on, but instead of a single humanoid robot (which I always considered to be a bad approach), they're small and specialized.
"In the developing world, 24,000 children under the age of five die every day from preventable causes like diarrhea contracted from unclean water." - UNICEF
That's 54 jumbo jets a day.
I also remain unimpressed by the ambition of his list. And by yours.
In my experience, Sand Hill Road does not want "frighteningly ambitious" startup ideas if substantial capital expenditure is involved. (In fairness, they are willing to hear those pitches – I guess that's something.)
> Now Steve is gone there's a vacuum we can all feel.
Pixar got funded only because Steve Jobs (Steve Jobs!) paid for it of pocket to the tune of $50 million total. It's Pixar that made him a billionaire (not Apple, as most people assume). How often does Steve Jobs invest in companies? Virtually never. But he knew (correctly) that Pixar was on to something.
I'm dealing with the Pixar bootstrap-problem at my own company, Fohr. Fohr is the live-action version of Pixar (photography, not animation, is what gets computerized), and requires $32 million in capital to do the process today on a feature film (well over half of that is for hardware - $2 million alone for electricity!).
Fohr is only constrained by capital – the R&D has already been done (it took nearly 13 years to develop the tech) – so you'd think Fohr would be ripe for funding. And you'd be dead wrong. There are no Steve Jobs left to pay for it.
The startup world today seems to only want tech innovation on the cheap, and that includes Paul Graham and all the rest.
So if I wanted to do PixActing like I wanted to breathe, my five year plan would be a) get VC funded for anything, b) achieve a modestly successful exit, and then c) recruit one similarly situated person and just shake the money tree. Without making disparaging comments about identifiable businesses, it is not a controversial observation that proven entrepreneurs with existing networks have vastly superior access to capital compared to first-time entrepreneurs with no network, independent of idea quality, target market, or execution ability.
$40 million is not a number that is unachievable in 2012. The password is just a bit different than for $200k, $700k, or $5 million.
You're right that VCs tend to be leery of the most ambitious ideas. That's another of the obstacles in your way if you pick one. But you shouldn't let your ambitions be limited by what VCs will fund.
(In any case you can trick them by only telling them about the initial few steps.)
Looks pretty interesting but I find myself having to guess what you're doing. It's important to be concise and reference things people already know ("Pixar") when you want to convey information quickly, but it's unclear what the value of the technology is in the 2 paragraphs I could find written about it. Everyone knows "Pixar" by name but I'm struggling to understand what you're doing. Is it animation software that maps photographs to the virtual world and renders it "almost-real"?
It's likely that even Steve Jobs would not have invested in Pixar if he knew how much it was going to cost to make it successful. He originally put up $10 million ($5 million for Lucas and $5 million to finance it). It became a money pit that either pride or faith compelled him to keep funding.
Sand Hill Road is just one source of money and money is the most fungible representation of wealth, so if they won't help you, go to somebody for whom 32 million is chump change or who are used to pay way more for a movie. Hollywood way be more receptive to your ideas (making a block buster isn't cheap and there is always a risk, so they should be used to taking them).
As an amateur historian, I found the Colombus bit a bit interesting, and probably more on-point than Graham might have even known. Columbus, his backers, and his detractors all accepted that the world was round. What they disagreed about was how big it was, and how far it would be to Asia by sailing West. Everybody, pretty much, by that point knew that the world was literally round (and flat only in stories). This was especially true in monastic and church circles which had known this for longer.
In other words they all agreed it was a great idea and an ambitious project that might succeed. They disagreed about what it would take to get there, and whether there might be obstacles in the way.
Seems like a very fitting metaphor for an ambitious startup.
Edit: For sources, you can start with "Heaven and Earth in the Middle Ages: The Physical World Before Columbus" by Rudolf Simek, which is a book uncommon in its level of insight. His description of Marco Polo's purported encounter with a unicorn had me laughing in both humor and amazement.
Simek's basic thesis was that Columbus's expedition was important historically because it blew away an important piece of medieval ethnographic thought--- once it became clear that the areas he had reached were not India, but were inhabited anyway, it doomed the Augustinian argument against the existence of inhabited continents beyond Africa, Asia, and Europe. This then paved the way for questioning the religious and classical basis for some aspects of the physical world, and lead in many ways to the Renaissance (though I think the failure of the Crusades and the translation of Arabic writings into Latin had a strong hand there too). The importance of Columbus's voyage about changing the way we think about our place on the world was still important. Another good point about ambitious startups?
In re #4, I'd suggest that your biggest hurdle isn't movie studios (as we often like to suggest here). It's Comcast. It's Time Warner Cable. It's AT&T. These companies exercise an oligopoly on most people's internet connectivity, TV UI and UX, DVR experience, etc. They also set the terms, with the networks and studios, for what you actually get to watch on demand. They pushed their crappy DVR onto the masses, effectively killing off the far more innovative and superior TiVo, because they offered their boxes at point-of-cable-hookup to consumers. They control so many strategic channels in the TV business, on both the B2B and B2C ends, that they're basically running the industry. (They were also the prime movers in the PIPA/SOPA legislation, and they'll be back with another attempt as surely as the sun rises in the East.)
Netflix, Apple, and Amazon look like compelling alternatives to the cable oligopoly. Unfortunately, studios are deathly afraid of handing over monopolistic control of their distribution to a single player like Netflix, so they're fighting with Netflix and trying to push their own alternative onto consumers (Ultraviolet). Meanwhile, they remain relatively oblivious to the real snakes in the grass (Comcast, et al.) -- an obliviousness that's going to get even worse, now that Comcast owns a major player in the production system.
To beat Hollywood isn't to beat the studios. To beat Hollywood is to beat cable. This isn't a war over content; this is a war over distribution. Technology vs. technology. Content producers will go wherever there's distribution to be found, and money to be made.
Here's another tip: I'm African, and I don't understand what you are talking about here. Maybe the next entertainment innovation should force global scale...
It's an interesting analysis but still the problem for any upstart distribution technology is to get content, it's a chicken-and-egg problem. I've worked in the past for content distribution technology company and the main issue was to get content, there were also specific issues to the technology chosen that made it sort-of-dead-end but I didn't see how it could get the content at relevant terms.
The technology to distribute the content is out there already, bittorrent showed the way and it is working at a fairly large scale. Any problem down the road technology-wise can be solved by some (non-trivial) amount of money and creativeness. It's not a technological problem.
The main trick is to get good content on a trivial distribution method. I've been thinking about this but I'm a technology guy and couldn't figure how to get the content.
I think personal health monitoring is probably the most important thing on that list. The thing that excited me the most when smartphones started becoming popular was the prospect that they could coordinate data collection from a number of sensors always collecting data - basic ones like Nike+, but perhaps also sensors measuring sleep, taking periodic bloodwork, etc. At the same time, perhaps you could automatically monitor personal behavior such as foods eaten.
Personal diagnostics would be an important use of that, but I think more importantly, with a very large public dataset of basic biometric data correlated with behavior data and medical results across a significant portion of the population, we could stop treating human health studies as bespoke one-offs put on at great expense and start treating them as data mining problems. You could begin to spot correlations between behaviors and results that are unintuitive given conventional wisdom. I think that the resulting burst of discoveries would be on par with any of history's great scientific revolutions.
There is a fundamental difference between a scientific study and data-mining.
Science is based on probability theory. Until we discover the "grand theory of everything", out other theories will be only approximate, and out experimental results not 100% predictable. Therefore, scientists consider a prediction as correct if the chance of predicting something at random is less than some probability, usually 10%, 5% or 1%.
However, for this to work, each study must be based on new data. If you use the same data to check e.g. 10 predictions, each of which has 10% chance of happening even if incorrect, you will in average confirm 1 of your predictions, even if all are incorrect!
Scott Adams (Dilbert guy) proposed this a few years ago as a way to win the Nobel Prize in Physiology or Medicine.
Being able to "tivo rewind real life" would be pretty amazing -- watch a bunch of things passively, record data, notice a spike in mortality, then find the common factors and stop the new plague (or the kids who found a Cobalt-60 source, or the pump infected with Cholera).
Interestingly I have a friend (a hematologist) that says that they actually try not to overscan for fear of finding something. There are plenty of cases where a scan reveals something that may not manifest itself into a noticeable issue in the lifetime of the patient but they then need to treat. Unfortunately often the treatments themselves are invasive (sometimes far more so that the issue itself might have been).
I guess that problem will diminish as advancements in medical diagnosis and treatment progress.
There's a definite opportunity to build a competitor to Apple and reach the hackers first: building a better PC for hackers to create new software built on Open Source and Web technologies.
The cracks are starting to show with using Mac OSX as a primary machine for hacking. It's got unix under the hood, but every successive release has become more consumer focused and less hacker friendly. The proprietary nature of developing native apps also turns off a lot of the great OSS hackers.
If you could get an all star team together with someone like Rahul Sood to design the hardware and someone like Miguel De Icaza to design the OS and developer APIs, you'd be well on your way to tackling this problem and building the next Apple. And this time, it could be a lot more open source friendly.
Do you think there is really a large number of hackers who are not satisfied by the Mac on the one hand, or Linux on a Thinkpad on the other, and who would buy into this new system?
OK, so you really just want to target hackers first and then move into the consumer space once you have some traction. But consumers are starting to move away from laptops and toward things like smartphones and tablets. Apple is moving in this direction and has a head start. If you start by making laptops, Apple's head start is just going to grow.
Better to figure out what is going to come after smartphones and tablets, and get there before Apple does. I don't see any particular reason to retrace Apple's steps and start by building a laptop.
Some companies I suspect might be following a long-term strategy similar to this already:
* Google, with the rumored heads-up display project based on Android
Do you know any hacker who knows about Raspberry Pi and isn't planning on buying one?
Radically undercutting existing platforms in price, but with comparable functionality enables totally new uses for general purpose computing devices. That's how to take on the Mac/PC business - not by hitting it head on.
Agreed. My next laptop is going to be a Linux one, and I'm going to pay for a high-quality hardware system... I can only hope the drivers are high quality.
I'd pay $2250 for a high-quality Linux laptop (roughly the price of my OSX laptop). :-)
"GMail is slow because Google can't afford to spend a lot on it. But people will pay for this. I'd have no problem paying $50 a month."
Ok. Number of Paul Grahams in the world times $600/year = ?
Most people on the web are ridiculously stingy. "I would pay for this" is a terrible way to think for an entrepreneur. Believing that what we think represents the masses is a rookie mistake.
I hate to stray into politics but my scary ideas revolve around public policy and the various actions people undertake in the public sphere that affect it. More specifically: Is it possible, by providing better tools for publishing and accessing information, to substantially improve public policy debates? Can we reduce the very large rewards for dishonesty and the use of disinformation?
This is the crux of the problem with our current political system, I think. It's not campaign finance, it's not religion, it's not disagreements about economics, foreign policy, security vs liberty (a lovely false dichotomy) or what have you. It is simply the fact that lies win and truth loses. Or, if that statement is not necessarily true, it is true in the current practice.
So, if you buy my premise, how can technology help? Isn't it a problem of human nature? You can't force people to be honest. You also can't force people to learn how to recognize dishonesty in spheres where they have not much competence. You can't impose good sense or decency.
But human nature is varied, and so maybe the seeming ascendancy of its more unfortunate aspects is situational. Maybe by improving the context and presentation of information they can be mitigated. Maybe technology can be used to recognize and reward honesty and to point out and discourage dishonesty. It hurts to think about, doesn't it? It does for me, because it is so hard, and that's what I took from pg essay. Granted, I may not be talking about problems to solve which would make you the next Google.
As an aside, I think that the utility of greater transparency of public actions (governmental or corporate) is already well-understood by many and much work is already being done in this direction so I am leaving out. But that doesn't mean there isn't room for new solutions there, as well.
The thing about replacing e-mail is that is isn't just a todo list, for many people it's just a receipt box - the thing I keep all my notifications that I bought stuff from amazon. For others, it's still the primary means of business communication.
My work e-mail is largely about communications, with a todo element to it and unfortunately some file storage too. My "home" e-mail is completely different. It's where I get my monthly statements for banks and investments and where my notifications go. When replacing e-mail you would need to service all these components of what e-mail is.
The thing that originally made e-mail so important was it's identity factor. That seems to have withered away as other services have replaced some components of what e-mail was for.
I would argue that e-mail needs to not be replaced, just reclaimed. My e-mail client (web or otherwise) should know that an e-mail in this case is actually just a twitter DM notification and be smart about how it presents that to me. It should know that something from Bank of America is probably something I want to keep, but something else from Bank of America is just marketing junk.
I haven't seen anything that is smart enough to do that on it's own. I don't want to have to deal with creating filters - it should just know. I would totally switch from gmail if this were out there.
- as I look at that old Palm IIIxe sitting in the cradle on my desk,
my mind swims in the ideas of all the databases on all of the devices everywhere all being in sync
3. replace universities
- yup, and recreate the free university of olde
4. internet drama
- indie drama titles (movies/shows/etc) on netflix / apple
- or just read the comments ;)
5. the next steve jobs - but I am still impressed by the iPad3 rollout and it's screen
- raspberry pi
6. bring back moore's law
- easy parallelism in software
- it's a compiler -- that's the hard part
- a compiler on the web as a web "service"
- an optimization marketplace: people in the machine doling out smart answers
7. ongoing diagnosis
- how about ongoing prevention? because cancer is a symptom too
- why limit yourself to 1000 years of ?barbaric? western medicine?
- why not look at all of humanity's history of medicine from all cultures?
8. tactics
- remember that columbus was a tyrant, and he didn't "discover" anything.
- start small
yup, the best plan is not to have one, and never make one, when it's a fait accompli then you announce the plan
If Microsoft : Google :: Google : Facebook, I'm not sure that the frightening startup idea here is to replace Google. Don't get me wrong, I've also started to see some cracks in the G edifice; and Facebook has definitely begun to set their agenda, but doesn't that mean search in general is already waning and that the next big thing will be whatever replaces Facebook?
Great essay though, lots to think about. I really like the anecdote about bolting an iMac to the wall as well. I still have a TV, but it's only purpose is to act as a large dumb monitor for my laptop, and I've been seeing a lot of this type of thing happening even among my non-hacker friends and family. I'd like to see an 'app-store' translation for drama as well, but it seems like tv / movies are not as amateur friendly to create as games. One person can develop a fun indie game, but it's nigh on impossible to create drama with a similarly small budget.
What aspiring drama writers / directors need are tools equivalent to game level editors to create their scenes without actors, cameras or studios. Packaged believable human CG characters may not be possible, but cartoon, animal, alien, etc. characters might be able to bridge the gap the way they do in video games and still tell a compelling story.
We're a YC company (www.post.fm) working on one of those ideas - email. We've encountered lots of headwinds as PG mentions, and we've managed to stay strong to our beliefs by keeping the team small and focused, and using our product ourselves to constantly remind us that what we're working on is better than Gmail.
It's not been easy admittedly, we didn't come up with some small idea that could grow into an email replacement over time, or some add-on to gmail to give us early traction, etc. We focused on replacing Gmail from day one. And that's no small feat, cause who wants to use a minimum viable email service?
We also realize it's a huge bet, and we may be wrong. But at least we're building something for ourselves, so we can't be too wrong, and that thought keeps us going.
Can't say I agree with "Email was not designed to be used the way we use it now. Email is not a messaging protocol. It's a todo list."
Email was designed to be the electronic version of a letter - an async messaging channel. Not some to-do list protocol. But with increased volumes managing all that mail became difficult (I'm sure celebrities still struggle to catch up with physical mail). That's the problem we want to solve, by letting algorithms and better user interfaces help you manage your mail.
A to-do list is something different in my view, but naturally closely related (and should be part of the same application). A piece of mail often prompts you to create an associated to-do item, but today this functionality isn't integrated so we rarely bother.
Sure IM, Twitter, and To-Do list apps chip away at some of email's use cases, just like instagram is doing with facebook, but we're confident that email can be just as good if done right.
Now I just have to finish it and avoid thinking about my idea for the google-search-killer that would be oh-so-easy to try out ;)
This all assumes conditions as usual. Which I think is a painfully mistaken notion. In an era of depleting fossil fuels, inferior ore quality of iron, copper and all the rare earth metals required to make modern electronics, water shortages and overpopulation, I doubt people will be worried about email overload in even 20 years. What's missing from this list are the real big issues of our time, such as quitting our reliance on fossil fuels, building a sustainable economy for the planet's resources, and creating a currency not solely based on debt expansion. In an economy with less energy surplus our problems will be more primitive than worrying about heart disease or faster computation or better search. Solving the shrinking energy surplus..., now that would be a really scary big startup of planet-wide implication.
"A New Search Engine": You don't need to compete head-on with the biggie(s) even if you want better search. You can vastly improve search by specializing, i.e. getting data that the biggie(s) don't yet have. Find something that is "universal" but where the necessary data is hard to obtain. Then innovate on how to obtain that data, rather than focus on how to search or match etc. Find a business model where you provide value for both those who generate/grant access to the data as well as those who use/search it. When you have that data, either build on it or sell it to one of the biggies.
"The Next Steve Jobs": Watching the TED talk by Cynthia Breazeal (http://www.youtube.com/watch?v=eAnHjuTQF3M&t=09m25s), it recently struck me that the next Apple will come from the robotics side. The engagement level and seamlessness you can achieve with the physical medium is on a completely different level from "devices". Even smartphones/apps require (at best) minimal cognitive facilities for interaction. Being able to I/O on a "reptilian brain" basis with body language, tone of voice, etc. could literally "change everything" by weaving intelligence into even the dumbest activity. People will not be able to do without such robots if they are well done.
If you want to take on a problem as big as the ones I've discussed, don't make a direct frontal attack on it. Don't say, for example, that you're going to replace email. If you do that you raise too many expectations. Your employees and investors will constantly be asking "are we there yet?"
This is critical. I have tried it the other way, and struggled for these very reasons.
I think the way to use these big ideas is not to try to identify a precise point in the future and then ask yourself how to get from here to there, like the popular image of a visionary. You'll be better off if you operate like Columbus and just head in a general westerly direction. Don't try to construct the future like a building, because your current blueprint is almost certainly mistaken. Start with something you know works, and when you expand, expand westward.
Intel has pretty good automatic parallelization (https://www.google.com/search?client=opera&rls=en&q=...) in their latest C++ compiler, but it's $1000+. I remember at a php conference they simply compiled php with their compiler versus gcc, and everything ran significantly faster in various benchmarks.
For a while I was working on automatic parallelization, and wrote plans, white papers, etc. but at some point was introduced to the current methods of automatic parallelization and saw that there are some pretty good solutions out there right now such as Intel's C++ compiler.
Ideally, everything would be compiled with something along those lines at which point the baseline for everything else would take advantage of multiple cores, at least in the simple to advanced cases without additional direction from the programmer.
After all, so long as you're not eval'ing you know the entire scope of the program and you can link things up as parallel independent queues. It requires more storage during compilation, and likely longer compilation times, but the performance result can be dramatic.
It's disconcerting that something like Intel's apparently wonderful automatic parallelization C++ compiler isn't more popular, even though it's demonstrably better performance-wise than anything else I've seen.
[+] [-] jballanc|14 years ago|reply
The Newcomen Engine has a fatal flaw: it cools the steam for the return stroke, losing energy to the latent heat of evaporation each time. James Watt discovers the latent heat of evaporation, and realizes that separating the condenser from the piston would improve efficiency. So let's go build some railroads, right? Not so fast. It would still be another 30 years (100 years from the invention of the Newcomen Engine) before railroads and ferry boats would be regularly powered by reciprocating steam engines.
What's the moral? For 100 years, vast leaps in technology came one after the other. In the process, the Laws of Thermodynamics were discovered and described. Many learned men stood around patting each other on the back at how successful, how inventive they were...at digging a black rock called "coal" from the ground.
But most people don't dig rock from the ground. Most people do travel from point A to point B on a fairly regular basis. The world changed when 100 years of technology left the mine shaft and the factory, and got people where they were going just a bit faster.
I'm convinced that computers are still at the Newcomen/Watt transition. We have a ways to go before the world truly changes.
[+] [-] waterside81|14 years ago|reply
These are largely first world problems. Here are some ambitious ideas:
- distributed power generation that's cheap enough and renewable enough so people in rural parts of sub-Saharan Africa don't have brown outs anymore.
- synthetic food generation a la star trek
- desalination that is cheap enough for a farmer in Mozambique to do himself
There are more, lots more. People outside the valley bubble have real problems.
[+] [-] wmeredith|14 years ago|reply
[+] [-] DennisP|14 years ago|reply
Another is about synthetic food...the presenter said his company could feed the entire world from an area the size of Rhode Island: http://www.youtube.com/watch?feature=player_embedded&v=r...!
[+] [-] Nervetattoo|14 years ago|reply
Education needs to catch up with the world in the first world, but if you do that in an async manner online then you can at the same time solve the problem with education in africa.
[+] [-] bh42222|14 years ago|reply
desalination that is cheap enough for a farmer in Mozambique to do himself
I am sorry, but you can not fix broken states with technology. Want to help Mozambique and sub-Saharan people? Find a way to turn their governments into well working ones. This is at once infinitely more difficult and also the only thing which will really solve the common developing world problems.
[+] [-] netcan|14 years ago|reply
Non consumption is traditionally a good place to start and I would suggest that high school & junior university level is the most disruptive place to start. Non-consumption of senior high school to junior university level education is something in great abundance in poor countries.
There is also probably a more incentive for potential students to play along in poor countries. A 9th-10th year dropout is more likely to be in that position because of access or soem other problem that technological innovation is good at dealing with. More importantly, the ROI on those 2-4 years of education is probably much higher.
If you were to go after job skills/ training as the point of attack (as opposed to general education) poor countries are also a great place to start. For a lot of skills there is demand at the bottom: bookkeeping, graphic design, programming, etc. Bringing a person making <$1-$2 per hour to a point where they can command a $3+ is fundamentally doable. That's a big incentive.
[+] [-] rmc|14 years ago|reply
[+] [-] Alex3917|14 years ago|reply
- No way to search for a scientific question and get a summary of the current scientific consensus or viewpoints on specific issues
- It's really hard to access academic journal articles online.
- Even when you can access journal articles, it's hard to know which ones to look in to answer your question. Sometimes it's hard to even know which field(s) your question falls under.
- Even if you vaguely know which field your question falls under, you don't necessarily know any of the vocabulary used by that field.
- No way to search by dependent and independent variables, confounding variables, etc.
- No way to sort articles by the quality of their methodology, the quality of the journal they were published in, the quality of the researchers, etc.
I know this isn't a product that more than 1% of the population would use, but if someone built it then maybe there are other things it could be used for.
[+] [-] dgallagher|14 years ago|reply
One thing I search for sometimes are code examples in a particular language. Search for something in C on Google and you end up with lots of stuff for C++, C#, etc... Github, with its large repository of public code, lets you filter by programming language, and is much better than Google in these cases.
Bing copies Google. DuckDuckGo returns different things than Google, but otherwise is a copy. There's no micro search engine for specific topics and sub-topics, outside of site-specific search. Market opportunity...
[+] [-] run4yourlives|14 years ago|reply
Where's the lunar base? The flying car? The personal robot? The cyborg? Meh, maybe I'm just getting old and grumpy. (To be fair I did like the other ones).
[+] [-] nl|14 years ago|reply
Search is another word for artificial intelligence.
Email means a universal way for people to communicate.
Lunar bases & flying cars aren't ambitious, they are just expensive.
Personal robots are here now in vertical spaces[1] and there are at least 200,000 cyborgs walking around now[2]. I think both these areas are worth working in, but I think you underestimate the ambition of a word like search.
[1] http://www.irobot.com/
[2] http://en.wikipedia.org/wiki/Cochlear_implant
[+] [-] pg|14 years ago|reply
I almost added a section on robots, but the talk was already long enough and I figured 7 was enough to illustrate my point that ambitious startup ideas are frightening.
[+] [-] staunch|14 years ago|reply
Look at Star Trek. Other than the spaceship itself the most impressive things are the technology the people use. The Siri-like computer interface, iPad-like computers, Skype-like communications technology, etc.
[+] [-] jakejake|14 years ago|reply
But, I remember when google was first released. I thought the search engine market was already full with yahoo and alta vista dominating at the time. Who knew?!
[+] [-] zem|14 years ago|reply
[+] [-] mcartyem|14 years ago|reply
[+] [-] icebraining|14 years ago|reply
And the personal robot is already being worked on, but instead of a single humanoid robot (which I always considered to be a bad approach), they're small and specialized.
[+] [-] ahoyhere|14 years ago|reply
"In the developing world, 24,000 children under the age of five die every day from preventable causes like diarrhea contracted from unclean water." - UNICEF
That's 54 jumbo jets a day.
I also remain unimpressed by the ambition of his list. And by yours.
[+] [-] erichocean|14 years ago|reply
> Now Steve is gone there's a vacuum we can all feel.
Pixar got funded only because Steve Jobs (Steve Jobs!) paid for it of pocket to the tune of $50 million total. It's Pixar that made him a billionaire (not Apple, as most people assume). How often does Steve Jobs invest in companies? Virtually never. But he knew (correctly) that Pixar was on to something.
I'm dealing with the Pixar bootstrap-problem at my own company, Fohr. Fohr is the live-action version of Pixar (photography, not animation, is what gets computerized), and requires $32 million in capital to do the process today on a feature film (well over half of that is for hardware - $2 million alone for electricity!).
Fohr is only constrained by capital – the R&D has already been done (it took nearly 13 years to develop the tech) – so you'd think Fohr would be ripe for funding. And you'd be dead wrong. There are no Steve Jobs left to pay for it.
The startup world today seems to only want tech innovation on the cheap, and that includes Paul Graham and all the rest.
[+] [-] patio11|14 years ago|reply
$40 million is not a number that is unachievable in 2012. The password is just a bit different than for $200k, $700k, or $5 million.
[+] [-] pg|14 years ago|reply
(In any case you can trick them by only telling them about the initial few steps.)
[+] [-] lusr|14 years ago|reply
[+] [-] staunch|14 years ago|reply
[+] [-] tomjen3|14 years ago|reply
[+] [-] ricardobeat|14 years ago|reply
The video on AngelList looks like just a rendering demo, and I can't find any other references.
[+] [-] einhverfr|14 years ago|reply
In other words they all agreed it was a great idea and an ambitious project that might succeed. They disagreed about what it would take to get there, and whether there might be obstacles in the way.
Seems like a very fitting metaphor for an ambitious startup.
Edit: For sources, you can start with "Heaven and Earth in the Middle Ages: The Physical World Before Columbus" by Rudolf Simek, which is a book uncommon in its level of insight. His description of Marco Polo's purported encounter with a unicorn had me laughing in both humor and amazement.
Simek's basic thesis was that Columbus's expedition was important historically because it blew away an important piece of medieval ethnographic thought--- once it became clear that the areas he had reached were not India, but were inhabited anyway, it doomed the Augustinian argument against the existence of inhabited continents beyond Africa, Asia, and Europe. This then paved the way for questioning the religious and classical basis for some aspects of the physical world, and lead in many ways to the Renaissance (though I think the failure of the Crusades and the translation of Arabic writings into Latin had a strong hand there too). The importance of Columbus's voyage about changing the way we think about our place on the world was still important. Another good point about ambitious startups?
[+] [-] jonnathanson|14 years ago|reply
Netflix, Apple, and Amazon look like compelling alternatives to the cable oligopoly. Unfortunately, studios are deathly afraid of handing over monopolistic control of their distribution to a single player like Netflix, so they're fighting with Netflix and trying to push their own alternative onto consumers (Ultraviolet). Meanwhile, they remain relatively oblivious to the real snakes in the grass (Comcast, et al.) -- an obliviousness that's going to get even worse, now that Comcast owns a major player in the production system.
To beat Hollywood isn't to beat the studios. To beat Hollywood is to beat cable. This isn't a war over content; this is a war over distribution. Technology vs. technology. Content producers will go wherever there's distribution to be found, and money to be made.
[+] [-] majani|14 years ago|reply
[+] [-] baruch|14 years ago|reply
The technology to distribute the content is out there already, bittorrent showed the way and it is working at a fairly large scale. Any problem down the road technology-wise can be solved by some (non-trivial) amount of money and creativeness. It's not a technological problem.
The main trick is to get good content on a trivial distribution method. I've been thinking about this but I'm a technology guy and couldn't figure how to get the content.
[+] [-] ericd|14 years ago|reply
Personal diagnostics would be an important use of that, but I think more importantly, with a very large public dataset of basic biometric data correlated with behavior data and medical results across a significant portion of the population, we could stop treating human health studies as bespoke one-offs put on at great expense and start treating them as data mining problems. You could begin to spot correlations between behaviors and results that are unintuitive given conventional wisdom. I think that the resulting burst of discoveries would be on par with any of history's great scientific revolutions.
[+] [-] tomp|14 years ago|reply
Science is based on probability theory. Until we discover the "grand theory of everything", out other theories will be only approximate, and out experimental results not 100% predictable. Therefore, scientists consider a prediction as correct if the chance of predicting something at random is less than some probability, usually 10%, 5% or 1%.
However, for this to work, each study must be based on new data. If you use the same data to check e.g. 10 predictions, each of which has 10% chance of happening even if incorrect, you will in average confirm 1 of your predictions, even if all are incorrect!
[+] [-] rdl|14 years ago|reply
Being able to "tivo rewind real life" would be pretty amazing -- watch a bunch of things passively, record data, notice a spike in mortality, then find the common factors and stop the new plague (or the kids who found a Cobalt-60 source, or the pump infected with Cholera).
[+] [-] aidos|14 years ago|reply
I guess that problem will diminish as advancements in medical diagnosis and treatment progress.
[+] [-] hkarthik|14 years ago|reply
The cracks are starting to show with using Mac OSX as a primary machine for hacking. It's got unix under the hood, but every successive release has become more consumer focused and less hacker friendly. The proprietary nature of developing native apps also turns off a lot of the great OSS hackers.
If you could get an all star team together with someone like Rahul Sood to design the hardware and someone like Miguel De Icaza to design the OS and developer APIs, you'd be well on your way to tackling this problem and building the next Apple. And this time, it could be a lot more open source friendly.
[+] [-] tesseract|14 years ago|reply
OK, so you really just want to target hackers first and then move into the consumer space once you have some traction. But consumers are starting to move away from laptops and toward things like smartphones and tablets. Apple is moving in this direction and has a head start. If you start by making laptops, Apple's head start is just going to grow.
Better to figure out what is going to come after smartphones and tablets, and get there before Apple does. I don't see any particular reason to retrace Apple's steps and start by building a laptop.
Some companies I suspect might be following a long-term strategy similar to this already:
* Google, with the rumored heads-up display project based on Android
* Jawbone
* Razer (in this case Apple is not the target)
[+] [-] nl|14 years ago|reply
Do you know any hacker who knows about Raspberry Pi and isn't planning on buying one?
Radically undercutting existing platforms in price, but with comparable functionality enables totally new uses for general purpose computing devices. That's how to take on the Mac/PC business - not by hitting it head on.
[+] [-] clavalle|14 years ago|reply
[+] [-] pnathan|14 years ago|reply
I'd pay $2250 for a high-quality Linux laptop (roughly the price of my OSX laptop). :-)
[+] [-] diego|14 years ago|reply
Ok. Number of Paul Grahams in the world times $600/year = ?
Most people on the web are ridiculously stingy. "I would pay for this" is a terrible way to think for an entrepreneur. Believing that what we think represents the masses is a rookie mistake.
[+] [-] aptwebapps|14 years ago|reply
I hate to stray into politics but my scary ideas revolve around public policy and the various actions people undertake in the public sphere that affect it. More specifically: Is it possible, by providing better tools for publishing and accessing information, to substantially improve public policy debates? Can we reduce the very large rewards for dishonesty and the use of disinformation?
This is the crux of the problem with our current political system, I think. It's not campaign finance, it's not religion, it's not disagreements about economics, foreign policy, security vs liberty (a lovely false dichotomy) or what have you. It is simply the fact that lies win and truth loses. Or, if that statement is not necessarily true, it is true in the current practice.
So, if you buy my premise, how can technology help? Isn't it a problem of human nature? You can't force people to be honest. You also can't force people to learn how to recognize dishonesty in spheres where they have not much competence. You can't impose good sense or decency.
But human nature is varied, and so maybe the seeming ascendancy of its more unfortunate aspects is situational. Maybe by improving the context and presentation of information they can be mitigated. Maybe technology can be used to recognize and reward honesty and to point out and discourage dishonesty. It hurts to think about, doesn't it? It does for me, because it is so hard, and that's what I took from pg essay. Granted, I may not be talking about problems to solve which would make you the next Google.
As an aside, I think that the utility of greater transparency of public actions (governmental or corporate) is already well-understood by many and much work is already being done in this direction so I am leaving out. But that doesn't mean there isn't room for new solutions there, as well.
[+] [-] rmassie|14 years ago|reply
My work e-mail is largely about communications, with a todo element to it and unfortunately some file storage too. My "home" e-mail is completely different. It's where I get my monthly statements for banks and investments and where my notifications go. When replacing e-mail you would need to service all these components of what e-mail is.
The thing that originally made e-mail so important was it's identity factor. That seems to have withered away as other services have replaced some components of what e-mail was for.
I would argue that e-mail needs to not be replaced, just reclaimed. My e-mail client (web or otherwise) should know that an e-mail in this case is actually just a twitter DM notification and be smart about how it presents that to me. It should know that something from Bank of America is probably something I want to keep, but something else from Bank of America is just marketing junk.
I haven't seen anything that is smart enough to do that on it's own. I don't want to have to deal with creating filters - it should just know. I would totally switch from gmail if this were out there.
[+] [-] mmphosis|14 years ago|reply
3. replace universities
4. internet drama 5. the next steve jobs - but I am still impressed by the iPad3 rollout and it's screen 6. bring back moore's law 7. ongoing diagnosis 8. tactics yup, the best plan is not to have one, and never make one, when it's a fait accompli then you announce the plan[+] [-] sawyer|14 years ago|reply
Great essay though, lots to think about. I really like the anecdote about bolting an iMac to the wall as well. I still have a TV, but it's only purpose is to act as a large dumb monitor for my laptop, and I've been seeing a lot of this type of thing happening even among my non-hacker friends and family. I'd like to see an 'app-store' translation for drama as well, but it seems like tv / movies are not as amateur friendly to create as games. One person can develop a fun indie game, but it's nigh on impossible to create drama with a similarly small budget.
What aspiring drama writers / directors need are tools equivalent to game level editors to create their scenes without actors, cameras or studios. Packaged believable human CG characters may not be possible, but cartoon, animal, alien, etc. characters might be able to bridge the gap the way they do in video games and still tell a compelling story.
[+] [-] akrymski|14 years ago|reply
It's not been easy admittedly, we didn't come up with some small idea that could grow into an email replacement over time, or some add-on to gmail to give us early traction, etc. We focused on replacing Gmail from day one. And that's no small feat, cause who wants to use a minimum viable email service?
We also realize it's a huge bet, and we may be wrong. But at least we're building something for ourselves, so we can't be too wrong, and that thought keeps us going.
Can't say I agree with "Email was not designed to be used the way we use it now. Email is not a messaging protocol. It's a todo list."
Email was designed to be the electronic version of a letter - an async messaging channel. Not some to-do list protocol. But with increased volumes managing all that mail became difficult (I'm sure celebrities still struggle to catch up with physical mail). That's the problem we want to solve, by letting algorithms and better user interfaces help you manage your mail.
A to-do list is something different in my view, but naturally closely related (and should be part of the same application). A piece of mail often prompts you to create an associated to-do item, but today this functionality isn't integrated so we rarely bother.
Sure IM, Twitter, and To-Do list apps chip away at some of email's use cases, just like instagram is doing with facebook, but we're confident that email can be just as good if done right.
Now I just have to finish it and avoid thinking about my idea for the google-search-killer that would be oh-so-easy to try out ;)
[+] [-] wolframarnold|14 years ago|reply
[+] [-] HSO|14 years ago|reply
"The Next Steve Jobs": Watching the TED talk by Cynthia Breazeal (http://www.youtube.com/watch?v=eAnHjuTQF3M&t=09m25s), it recently struck me that the next Apple will come from the robotics side. The engagement level and seamlessness you can achieve with the physical medium is on a completely different level from "devices". Even smartphones/apps require (at best) minimal cognitive facilities for interaction. Being able to I/O on a "reptilian brain" basis with body language, tone of voice, etc. could literally "change everything" by weaving intelligence into even the dumbest activity. People will not be able to do without such robots if they are well done.
[+] [-] ajju|14 years ago|reply
This is critical. I have tried it the other way, and struggled for these very reasons.
I think the way to use these big ideas is not to try to identify a precise point in the future and then ask yourself how to get from here to there, like the popular image of a visionary. You'll be better off if you operate like Columbus and just head in a general westerly direction. Don't try to construct the future like a building, because your current blueprint is almost certainly mistaken. Start with something you know works, and when you expand, expand westward.
Eat small morsels, chew well!
[+] [-] underwater|14 years ago|reply
Can we please stop using that word? It trivializes the disorder and encourages the stereotype that engineers should be socially awkward.
[+] [-] asnyder|14 years ago|reply
For a while I was working on automatic parallelization, and wrote plans, white papers, etc. but at some point was introduced to the current methods of automatic parallelization and saw that there are some pretty good solutions out there right now such as Intel's C++ compiler.
Ideally, everything would be compiled with something along those lines at which point the baseline for everything else would take advantage of multiple cores, at least in the simple to advanced cases without additional direction from the programmer.
After all, so long as you're not eval'ing you know the entire scope of the program and you can link things up as parallel independent queues. It requires more storage during compilation, and likely longer compilation times, but the performance result can be dramatic.
It's disconcerting that something like Intel's apparently wonderful automatic parallelization C++ compiler isn't more popular, even though it's demonstrably better performance-wise than anything else I've seen.