HN user

writimov

46 karma
Posts0
Comments27
View on HN
No posts found.

One other thing I haven't seen anyone tackle is to evaluate 'best' durable via first principles equations and engineering.

Ex of a fridge:

* Define the physics of how the product performs over its life: physics of door opening and effort to lift the door, compressor energy to turn on, vibration life cycles and wear, power converter wear, etc.

* Define theoretical best performance for the product for each use case in fundamental physics.

* Get performance data for real test units.

* OR if testing real units is too expensive and time consuming. Put modeled fake units in a physics simulation engine (like https://docs.unity3d.com/Manual/PhysicsSection.html) and stochastically seed millions of virtual units with minor differences and virtually test to failure.

This would result in a much stronger measure of 'best' durable, that then, the crowd-sourced reviews would, hopefully, reiterate.

Good prototype. To achieve your goal of best products in categories and subcategories, you will need a few things:

* Universal equations that can be applied to all products or major categories of products for what is 'best' durable for long term use, including adjustments for value/price. Have to evaluate for self-repair vs pro repair. Availability and price of spare parts. Account for geographic location.

* To help sell the 'best' recommendation for subcategories of 'best', you could show other products that were close but are not the best. But to help users see 'best' in specific categories (ex. not best durable fridge, but best quietest fridge) you will need many custom filters. Some ideas to copy are the extensive filters on Digikey for say, an alum. electrolytic capacitor: https://www.digikey.com/en/products/filter/aluminum-electrol... or on McMasterCarr for a screw: https://www.mcmaster.com/screws

* A good test of 'best' durable in my mind, is answering the question: Would this product be the one chosen to go with the Mars colony ships? Is it fairly cheap, durable in all environments, easy to repair, easy to make spare parts, lasts a long time, low maintenance, etc?

Q: How it can get so much better?

Ans: From the most recent full presentation (pretty dated at this point Dec 2018): https://www.youtube.com/watch?v=nSIzsMlwMUY

From 10:22mins in the video: * Increase speed: automated segment erection and logistics; simultaneous mining and installation of reinforcement segments; 3X increase in power for TBM; continuous dirt removal; modified cutter design. Probably leads to 10X faster.

* 13:12mins: Decrease costs: onsite integration, reduced diameter of tunnels, electric locomotive, sell bricks. 15% of costs is dirt removal. Selling bricks for $0.10ea pays for dirt removal. 15% savings right there. (Price comparison: bricks cost $0.25 retail at Home Depot)

So the test tunnel cost $10mil/mi but can only put through 1/4 the area. So to balance against existing tunnels at $1bil/mi lets round up to say equivalent throughput for mass is $50mi/mile - assumes people throughput happens at the same speed of current mass transit - 45mph. Hyperloop tunnels will go much faster than that. But even non-hyperloop electric cars can go 90mph with self-driving - so double speed = double mass transfer and half mass_transfer cost per mile per time unit.

So kinda apples to kinda apples comparison of the very first tunnel Boring Co made is $50mil/mi versus $1bil/mil in LA for same mass throughput at same speed. They are already 1/20th the cost. Even if I am way off, by a factor of 10, the Boring Co is already HALF the cost of existing tunnels with their very first attempt in Dec 2018. The next update will put them even further along.

Did a site search "boring site:eschatonblog.com" and read through top 30 or so hits. Didn't see any arguments that Boring Company is fundamentally wrong and can't succeed through the limitations of physics.

Valid criticisms:

* Boring Co is moving pretty slowly.

* There are specific problems that have not been solved yet related to station design, the mini bus people mover vehicle (12 people) has not been made yet, self-driving cars aren't real yet, the actual Boring machines are only on version 3, the automated tunnel wall mfg and assembly machinery has not yet been built.

I agree that there are many assumptions (which could be incorrect) that need to be solved for the city to city test to succeed:

* Cheaper Station design for passenger and cargo load / offload.

* 12 person walk on / off bus mover vehicle design, manufacture.

* Self driving.

* Faster elevators and probably bulk elevators to move vehicles between levels of tunnels.

* Faster Boring machines

* Tunnel wall assembly machines from dug out dirt.

* Automated tunnel wall assembly.

The first solutions to these problems will probably be pretty crappy. But they will improve rapidly.

It's not clear how much of the Prufrock (3rd version of TBM) was built in house by the Boring Co. https://cleantechnica.com/2020/03/01/prufrock-next-generatio...

Godot (1X speed) was a straight up purchase of existing tech. Linestorm (2X speed) was partially upgraded from existing machinery from other companies by Boring Co. Prufrock (2->10X speed?) is the latest and version 3.

My guess is they need to get up to version 10 or more for a chance to beat Gary the snail. They still have a long way to go before they have their own machines 100% built by Boring Company that are much cheaper and faster.

Also, it doesn't look like there is much progress yet on the brick making machines. Or the automated tunnel assembly from bricks made from local dirt. The ideal system has definitely not been assembled yet: Boring machine -> dirt into wall sections -> wall sections assembled into tunnel wall.

So yes, there is a long way to go yet. But these are solvable problems that the team is working on.

The selling point: If you want to live in a big city for its job and cultural opportunities but can't afford it, then this option is 10X cheaper.

That makes this option very attractive. Especially if you choose a town franchise that fits who you are. You could join the Tech Town if you are into technology. Or Scientist town if you want to do University research. Or the Middle Ages town if you want to ride a horse and a visit a Blacksmith.

And if you change your mind on who you are in a few year, and one of the other 20 towns near a big city look attractive, you could move there. Say you want to party in Mini-Vegas town for your early 20s but then move to University town in your early 30s to get educated. That could be possible.

Overall point: There is a potential solution to the housing crisis for almost every city. It is too early to tell if it will work. But it looks promising.

Steps:

1) Build electric cars (exhaust gases in tunnels kill people). (Completed 2020 - Tesla)

2) Build tunnels that are 1/10th to 1/1000th cheaper than current technology. (In Progress - Boring Company)

3) Build tunnels that connect cities and the infrastructure to make city to city commutes through a tunnel but walking on foot using tunnel bus-like vehicles, easy, cheap and safe. (In Progress - Boring Company)

4) Build a test town of 1k-10k condos and connect to a major city with Boring Company tunnels. Initial cost of a test would be $500mil to $5bil. (Not started - ?)

5) Reduce the cost of building towns through 10X or better infrastructure improvements and develop a franchisable business model around towns.

6) Franchise the towns.

Result: Housing supply goes way up near where people want it right now. NIMBYs can't stop it. Investors may lose their shirts or make massive profits.

Advantages over other solutions: This only works if the business model works out. It leaves the politics of land-use within cities almost completely out of the picture.

Disadvantages: Its never been done before. It will cost a LOT of money to test it. The investors must be in it for the longish haul: 5-10 years at least.

Agreed. I don't have all the numbers worked out but rough calculations show that it is doable and even profitable within an order of magnitude or so.

Once the proof of a city to city test tunnel is completed, probably this one early next year: https://www.teslarati.com/boring-company-secret-tunnel-adela..., then the basic business model of the tunnels can be tested.

After that, it may be time to work out the details of building an ultra dense city on super cheap land and try to reduce the costs of construction and maintenance of city infrastructure: water treatment etc.

The selling point of a $200k condo with access to pools, gyms, etc with all the basic stores like dry cleaners, grocery store, restaurants a short walk away in your town/city AND a 10-25min tunnel ride to the big city may be hugely attractive for many people. Plus if you want to go ATVing or cherry picking or horseback riding or camping, you can take a 10min trip from your town right into the countryside because you will be there already, in the countryside.

If an average restaurant worker or entry level school teacher works in San Francisco right now, they may earn say $60k-$90k and pay 33% taxes - max $30k. Factor in rent at $30k (1bedroom is $2500/mn), and basic living expenses of $15k and that person is saving max $15k/year. The average condo in San Francisco is $1.2mil. So it would take around 8 years to save for 10% down payment or 16 years to save for a 20% down payment.

Contrast that with a condo for $200k a 20min commute away. You would have a 20% down payment in 3 years.

It takes longer than 20mins to drive across the city without traffic right now. During busy times it can 60 minutes to drive from the Golden Gate bridge to SOMA. You can literally walk faster than cars in San Francisco.

Agreed. The cost for underwater tunnels will be much higher - especially initially. Let's say they are 100X higher cost at $100mil/mile - that is still 1/30th the cost of a current tunnel in San Francisco at $3bil/mile. But the Boring Company is already designing for air tight tunnels: 'vacuum-sealed'.

"TBC’s current tunnels are designed and built in preparation for their eventual transition to Hyperloop."

"Hyperloop is an ultra-high-speed underground public transportation system in which passengers are transported at 600+ mph within a vacuum-sealed tunnel. Whereas Loop is used for shorter intra-city routes, Hyperloop will be used for longer inter-city routes."

https://www.boringcompany.com/faq

These guys have some good videos on Boring Company tunnels. https://www.youtube.com/watch?v=VGG1nWzP-Oc&list=PLDn5j59RTW...

The summary: the hard part of tunnels and people was what to do with exhaust gases. That's solved with electric cars. The tunnels will probably pay for themselves or at least break even.

In 5-10 years, its likely that most major freeway projects will have a very difficult time getting funding as tunnels will be cheaper to build and much much much cheaper to maintain than roadways. And existing investments in tollroads or bonds for roads will be selling for pennies on the dollar.

The cost per mile may even get as low as $100k/mile. At that price point, individuals could be drilling their own private tunnels.

As a comparison against tunnel prices now, when Google Fiber tried to lay fiber optic cable in San Francisco a few years ago using shallow trenches and not full tunnels, the full cost was estimated to be in the $100's of millions for the backbone and low billions including all house hookups. https://sfbos.org/sites/default/files/FileCenter/Documents/5...

And San Francisco is a tiny city: 7 miles by 7 miles. Once the Boring Company gets prices down to $1mil/mile, you could make the same Fiber backbone with 3 tunnels N-S and 3 tunnels E-W for 6x7 = $42million - a little more than 1/10th the quoted cost.

More comparisons: the Alaska Way Viaduct project (SR 99) in Washington State cost >$3billion and was only 2 miles long and took 12 years. The cost was $1bil/mile - so Boring Company will be 1000 times cheaper. https://www.wsdot.wa.gov/projects/viaduct/budget

The San Francisco BART extension cost $3bil/mile!! https://sf.curbed.com/2018/6/18/17464616/bay-area-subway-tra...

The solution is to build dense housing (min. 20 story towers) with mixed commercial in the lower levels on cheap land (<1/100th the cost of city land) within 50 miles of cities and then run Boring Company tunnels to your new town at a cost of $1mil/mile. Commute time of 10-25mins. Total cost is roughly $1.8bil for a town of 30,000. Each of the condos goes for $150k. That's 12% of the price of a condo in San Francisco. Also could build down underground 10-15 stories. Unsolved problem is low-cost infrastructure (water, sewage, power). Whatever couldn't be built on site could be piped in through through additional tunnels.

This would completely change housing prices in cities. If SF has around 700k residents and say another 700k people who want to live here or commute in each week, it would take 23 of these towns to accommodate demand: $46bil investment that would probably payout 25% within 5 years of construction start with all units sold, commercial leasing, naming rights, etc. (Large assumption is people would want to live in these towns). San Francisco housing prices would probably drop at least 25% in 5 years.

Just like Matatu's of Kenya (http://infinitedictionary.com/blog/wp-content/uploads/2016/0...) each town would be known for its thing: TownA is known for its Opera House, TownB is known for its Basketball Team, TownC is known for its amazing Foodie Scene, etc.

A successful 'Town' model could even be franchised like Burger King, repeated all over the USA or World within 50 miles of existing major cities. If you were the owner of a successful franchise, and optimized your margins (solving the electricity problem with solar panels or a cheap nuclear micro reactor), you could have 200% returns on each $1.5bil project. Even a couple of towns completing each year could be worth billions. And they would probably be less complicated to build than the average Tesla Terafactory, once the first few had been built.

YC has been trying a bunch of things new to them. Not sure how they are working out yet:

* Supporting more hardware specific companies.

* Supporting more biomed device tech companies.

* Scaling online with startup school.

* Trying to scale the number of companies who get accepted into the program.

There is still a lot to figure out in the 'nurture small/young companies' space. That said, each country has its own unique challenges.

If you want to use it for machine learning or VR/AR then it should ideally have a separate GPU board. Also nice to have: upgradable RAM, upgradable solid state drive bay.

Great product! But I almost quit your signup process. Forcing me to answer survey questions about my field and give my company name before I can see what your software can do makes me not want to sign up.

What does this do to the company's financial future? If there are thousands more of these lawsuits coming and the legal precedent has been set, and this one verdict took 40% of Bayer's cash position away, how can they survive the next 12 months? Are they bankrupt already if even 10% of these lawsuits go through with the same verdict? https://seekingalpha.com/symbol/BAYRY/balance-sheet#figure_t... Even if the verdict is slashed to 1/10th $100mil and they win 90% of their lawsuits, they will still lose 1,000. That's 1k X $100mil = $100bil in losses. Even cutting this by 50% they are still out of business.

I've used a Google Glass, Oculus Rift, Cardboard VR systems and seen a demo of Oculus Rift. Is the Lenovo ThinkReality a real competitor? The market hasn't picked up VR for gaming and AR has flopped so far. How about high end systems for engineering and design. There could be a case where engineering speed and collaboration could increase from using a few of these systems at any company doing hardware design.

Fantastic product! Really great that you have been able to help people become empowered for themselves.

Scale: Do you need help reducing the BOM costs? You could go for economies of scale (and reuse 75% of the same BOM) and build a variant for adult rehabilitation for injured soldiers, car accident victims or others with leg injuries.

Or how about a variant that allows for anyone to use to strengthen imbalances in muscle training/development? It could be gym equipment that you get on, it takes measurements of your leg strength by putting you through its paces, then it gives you a customized workout to balance your muscles for optimum strength. These exercises for imbalance measurement aren't an exact fit to the machine but can give you an idea: https://dailyburn.com/life/fitness/muscle-imbalances-functio...

I've spoken with multi-generation football family coaches, players who say that the sport is headed for an end. I'm not sure I believe that.

It comes down to whether a safe helmet can be made or whether a CTE treatment can be created. (Both of those seem like very difficult goals.)

The number of participants in the kids leagues has been declining for a while: https://www.statista.com/statistics/191658/participants-in-t... What parent would allow their high schooler to play if there is a 21% chance of lifelong disability, loss of earnings potential, etc.?

What can the average person do to help the oceans that is easy to understand, everyone can do themselves and the government can sponsor with money?

I'm thinking of something like planting trees: order a free tree (government paid) in the mail, pick a spot and plant it. * https://www.weforum.org/agenda/2018/07/pakistan-s-billion-tr... * https://news.globallandscapesforum.org/28687/africas-great-g...

Can we do the equivalent and plant a kelp bed or the equivalent for different ocean conditions? https://www.montereybayaquarium.org/animal-guide/plants-and-...

I'm having difficulty understanding how to use this. Is there a demo / example to view? I'm currently managing a team using Google Analytics, Facebook Pixel and Hotjar. How can I benefit from Avo?

From what I understand, I can use Avo to setup flows that I want to watch (like a sales conversion funnel) and make sure that the analytics tags are properly implemented across all pages of the funnel. Is that correct?

In the case that the sales funnel changes and a new page is added, do I need to remember to update Avo with the change?

This is the question I was really getting at: where is the site where I can go help beyond the ways that most people already know about? Where I can I volunteer my services to fix a posted problem with MikroTik routers being insecure? Where is the site with a giant list of problems and descriptions of the people needed to solve them? And matching algorithms to put the two together?

More to the point: where can someone who has a totally unique skillset help solve a specific problem?

The assumption I'm making (which may be faulty) is that everyone has a unique and valuable skillset that may only apply to specific problems. (Like for example, I know nothing about B list celebrities from 1950s Hollywood. But if someone had a problem to solve that involved knowledge of that period, they could post the problem and match it to a profile of a Movie professor, or just a regular person who happens to know alot about that subject. The technical complexities of proving someone knows what they are talking about and their authority can be trusted, basically filtering spam contributions, is the biggest technical challenge. But it seems like new tools like machine learning and neural nets could help us here.)

We all know we can help do basic low-skilled stuff: donate money, volunteer at homeless shelter, build houses habitat for humanity, volunteer at local garden, trash cleanup etc.

What website do I go to that collects all of these ways to help and more?

For ex, what website do I go to for helping in these scenarios:

1) I understand politics and want to work on bills in various countries to stop climate change?

2) I understand biology and want to cure the algae bloom I saw in my local lake yesterday?

3) I understand the physics of mechanical design and want to help fix a design flaw in a pair of garden shears that keeps cutting my skin between my thumb and first finger?

Or approach it the other way around: what site can I go to, register my interests and skills and get assigned to existing projects that will change the world? Where I can help anytime I have free time. Or get an assignment in my email?

Scenarios:

1) I sign up on the site saying I have a biology degree and live in Napa Valley, CA and can do environmental stuff. I get assigned to take Cesium 137 readings to followup on the Fukushima disaster in wine country at various GPS coordinates and get sent a geiger counter, instructions, and training.

2) I have Crohns disease but am willing to wear a data collection monitor on my body and send the data to multiple companies developing new medicines or treatments.

3) I care about elephants and can volunteer to take a shift watching through the eyes of an automated drone that scans for poachers on the other side of the planet?

There are plenty of ways to do basic help in a semi-organized manner to treat the symptoms of chronic human problems. There is no site that I've seen where solving the largest problems of humanity can be crowdsourced.

Where I can I volunteer my services to help design water desalinization plants to make them 100X cheaper?

This particular effort seems to be a mix of fun, braggadocio, and altruism. Could this sort of thing be organized with a social network and a list of tasks/problems using a tool like Trello or Jira but for solving any problem? The result would be anyone in the world could help/volunteer to fix real problems with free time. Use: 1) Problem is posted 2) Investigated and confirmed to be real 3) Volunteers start to fix and swap solutions 4) Extra people are recruited as necessary 5) Problem is solved and wrapped up. This could be applied from everything from MongoDB security issues (https://snyk.io/blog/mongodb-hack-and-secure-defaults) to cleaning up neighborhood pollution (http://www.chicagotribune.com/news/local/breaking/ct-chicago...) Thoughts? Does this exist already?

Reverse supply chain disassembly into raw resources for reuse is not being done in any meaningful way except for a glass, some chips, paper, precious metals and a few other things. We really need a large upgrade in the detection, cleaning, sorting, disassembly, prep for reuse and bulk source materials creation of human waste streams (garbage and recycling).

It starts with detection and sorting robots (one example: https://www.amprobotics.com). Add sensors to tell organic materials from plastics, sort into cleaning. Once cleaned use sensors to detect one type of plastic from another, then sort plastic types.

If one type of plastic is bound to another with a metal screw, use a disassembly robot to get the screw out.

Once sorted, use standard techniques to turn single plastic parts into bulk plastic materials ready for reuse. (Ex. of do-it-yourself version for HDPE https://www.instructables.com/id/Making-Blocks-out-of-HDPE-m...)

Using machine learning and the latest in robotics, this entire cascade could be accomplished but the business model is broken. The machinery would be too expensive to pay back with the value of the recycled materials. Instead we need to value the effort of keeping these pollutants out of the general environment. Carbon credits are a great example of how this value can be created.

Create a plastics pollution credits marketplace where you get paid to keep plastics out of landfills. Now maybe the robots, sensors, and recycling technologies are cost effective.

Could be useful to apply machine learning or a neural net to see patterns here. Examples: * There are many new words being generated in spacecraft design. Conclusions: this part of engineering and science is advancing rapidly. * The source of many new words come from patent filings. Conclusion: a new industry is being born in the systems biology space. The trends could be used to drive investment decisions, what to study or what type of startup to join or found.

Glad Anne Wojcicki, CoveredCA and Legal Aid were able to step in and help. There are some specific problems highlighted in the article that could be fixed by software startups:

1) Problem: Difficulty to identify a Doctor for a second opinion. a) Possible solution: software that allows you to input the medical term (or possibly the medicare code for a disease or combos of diseases?) and find all the second opinion doctors located near you. The software needs things like sort by reputation or reviews, price estimates, speed to get second opinion, time window (with a button to click for life-threatening emergency). Also, why be local? Through telemedicine you could video chat with your second opinion after digitally transferring records.

2) Problem: Too long to transfer medical records 5-10 days instead of 10mins. a) Possible solution: HIPAA and GDPR compliant digital medical health platform that works with all hospitals. (Very difficult to solve this because of non-standardization.) b) Secondary solution: Biometric authorization (fingerprint and voiceprint ID or DNA scan) and quick digital download of medical record in plain text. Must complete in less than 10mins even with slower bandwidth.

3) Problem: Didn't have health insurance. Hospital was looking at a potentially very large unpaid bill. a) Single payer government health insurance is the best solution. b) SW Tech company solution is quick instant insurance signup with government or charity guarantee for some percentage of bill. The rest of the bill is dumped into CoveredCA or something else. And the patient is enrolled for 1 year after the fact and must pay premiums. ? There are probably better solutions.

4) Problem: Must rely on humans. a) IBM Watson for disease diagnosis and treatment recommendations combined with additional neural net or machine learning technologies could be used to created automated second opinions. Feed all the raw data into the system to determine the outcomes and probability of recovery for all treatment options.

Its interesting. The article talks about the energy expended to do work (moving from place to place) and compares different types of human powered movement (running, biking, swimming). Then it dives into efficiency comparing one person racing a commuter train.

Side note: It would be cool to see efficiency of source of energy as well. Bananas eaten through a human digestive system vs diesel locomotives.

This could turn into a new metric for athletes - power output during certain types of physical activity. Ex, did you improve from 4.1W/kg to 4.3W/kg for a certain activity? What did you do to get there?

The answer might not just be about lost fat weight and increased muscle weight but more along the lines of improved form efficiency. Ex, for cycling, I improved by form so my knees don't move out so wide (kept in by 1cm) during the stroke do my drag coefficient from 0.88 to 0.84 and I become more efficient. https://www.cyclingpowerlab.com/CyclingAerodynamics.aspx You could really dive into this as a pro athlete. Just like IQ can be broken into sub-measures (overall IQ broken into spacial reasoning, puzzling ability, etc). Start with EQ (Energy Quotient) for a given activity then break it down into specifics.

Example 1) Michael Phelps has a swimming style that is more efficient than other swimmers. So his swimming EQ would be very high. And you could break it down into specific moves that make it high (ex his flipturn is fast, he breathes fewer times and that is fast, his butterfly kicks are better than usual) https://www.yourswimlog.com/michael-phelps-numbers/. Then train the specifics to get your EQ higher.

Example 2) Bode Miller skier with his upper body and arms all over the place but his lower body locked into the fastest way down the hill. His VMG, velocity made good, towards his goal is super high. So if we measured his energy expenditure and change in kinetic and potential energy coming down a hill, we could come up with a EQ number based on a Lagrangian. If other skiers wanted to match his lower body style, we could study their own Lagrangians and optimize for better EQ. Then those skiers could train with exercises to keep their EQs high by focusing on specific moves to perfect, better reading of the hill terrain, strengthening specific muscles, etc. That's the breakdown.