Tom Quisel

The Future of Agriculture

Wed, 09 Nov 2022 00:00:00 +0000

Crafting the job candidate experience

Tue, 23 Aug 2022 00:00:00 +0000

Happy job seekers on their way

Top job candidates receive many offers as they look for their next opportunity. Not to worry—your company can stand out from the pack by providing an outstanding experience throughout the interview process. As salaries and bidding wars over candidates heat up in the current high-inflation environment, differentiating your company is more important than ever.

Executed well, a great experience will markedly increase your job offer acceptance rate. That means you need to interview fewer candidates, and you can spend more time on each qualified candidate to provide an even better experience.

I’ve designed and iterated on dozens of open roles at top tech companies. Here are 7 tips I’ve learned to help you create that stellar candidate experience.

1. Move quickly

Great candidates will have lots of offers. The longer you take to interview them, the more likely they’ll accept an offer from a competitor who gets there first. Quickly sending an offer to a candidate sends two implicit signals:

You’re excited about them
Your company is well organized and can move decisively

Both will help win over the candidate. Challenge: can you get from the job application to offer within two weeks?

2. … and also be thorough

The goal is a quick yet thorough interview process that cuts out any unnecessary delays while still collecting robust information in each of the core areas you consider in your hiring decision. So take the time to design a thorough set of interview questions.

A strong candidate will be aware of the level of your interviews as they’re forming an opinion of your company. If the questions are too easy, are irrelevant, or fail to cover key topics, a candidate may conclude that the company is full of B and C players. After all, they can tell that the interview isn’t thorough enough to filter out lower-competence candidates. Conversely, a well-designed set of interview questions sends the message that your company only hires the best and is high-functioning enough to craft a quality interview.

3. Paint a clear picture of the interview process

Clearly communicate the interview process up front. This includes the steps, timeline, and flavor of the interview questions. The goal is for candidates to have all the information they need to put their best foot forward and to avoid surprises that knock them off balance. People do their best, most creative work when they’re at ease, and that’s what you really want to see in candidate interviews. The candidate will all be happiest when they’ve been able to show you their best. Great experience? Check.

4. Form a real human connection

From the very first contact, you want your candidates to feel respected and appreciated. This starts early on in an initial conversation with a recruiter who cares about your company mission, communicates it clearly, and begins to form a personal relationship with the candidate. It’s fantastic for your recruiter to position themselves as the candidate’s ally, ready to guide them through the interview process and look out for their needs.

Once the candidate has passed an initial skills screen, give them a chance to talk informally with their hiring manager. A candidate’s relationship with their prospective manager can be make or break as they’re deciding on their next career move. A low-pressure conversation gets this started on the right foot while making it clear to the candidate that you really value them.

As the candidate gets to meet more of your team during interviews, include ways for them to get to know members of your team personally. Inviting the candidate to lunch or drinks with the team is great if you’re in-person. One of my favorite interviews included a game of Settlers of Catan with the team. Games and other fun activities can be a great way to break the ice and form that connection.

5. Be transparent about your company

Give candidates a good 5–10 minutes to ask questions at the end of each interview session. It’s smart to have your interviewers prepare by thinking about how they’ll answer common questions, but make sure they’re answering questions honestly and being open about both your company’s strengths as well as areas for improvement.

6. Have a diverse interview panel

Ensure that your interview panel is diverse, represents different kinds of people at your company, and includes people whose lived experiences are relatable for the candidate. Not only will this lead to better hiring decisions, it’ll help your candidate feel more at home as they interview. The best way to make it clear to top diverse candidates that they have strong career growth potential at your company is to prove it. Have good representation in your hiring managers and leadership team.

7. Keep it comfy

Pillows on the interview couch—a nice touch

Take some time to design your interviews to be a pleasant experience:

Add time for bio-breaks and keep candidates well fed and hydrated.
Avoid more than two interviewers in one session.

For software engineering coding interviews, replace whiteboard coding with laptop coding. Ideally the candidate can use their own laptop or a familiar development environment.

I also recommend avoiding pressure interviews. There’s good evidence that they reduce creativity, they’re not fun, and they’re rarely realistic. If your company is run well, your team will spend most of their time working on projects with a deadline longer than an hour.

One enjoyable interview strategy involves the candidate and the interviewer working together to solve a problem, with the interviewer providing the framework for the candidate to make progress. If it’s done well, it creates a collaborative, energizing experience.

A great experience

Taking these 7 steps to craft a great candidate experience will increase your offer acceptance rate. And if you’re creating an extra-awesome candidate experience, you’ll find that even some candidates that you ultimately reject will refer your company to their friends. What better way to create a company culture and reputation that you can be proud of?

Navigating Disagreements as a Leader

Thu, 16 Jun 2022 00:00:00 +0000

Don’t be these birds.

I started leading the engineering team at OkCupid early in my career at the age of 24, after working for 3 years as a Software Engineer. It was an incredible opportunity to be given so much responsibility, but it came with many new challenges. Disagreements were new territory for me, and I found a variety of conversations ranging from giving feedback to salary negotiations to disagreeing about software design pretty uncomfortable.

In one memorable case, OkCupid’s databases were rapidly filling up and slowing down as the app grew. We had to act quickly to avoid a total meltdown. Instead, I ended up locked in a weeks-long disagreement with my head of infrastructure about the best approach. In my mind I knew I was right, but I just couldn’t get him to budge. I carefully explained the logic to him in depth. I pointed to the data, but I just couldn’t make progress. He explained his approach, but I just didn’t think it met the requirements. In the end I pulled rank and made the decision for him, a move that would have a lasting negative impact on our relationship and go down on my mental scorecard as a serious leadership failure on my part.

Looking back, I was tragically naive. The conversation wasn’t really about finding the best technical decision, it was about meeting deeper needs in the human psyche.

Here’s the advice I wish I’d heard as a new leader.

It’s easier if you have a good relationship

All the techniques in the world for having a productive difficult conversation won’t matter if the other person doesn’t feel safe and respected. By the same token, if the other person feels kinship with you, they’ll be more likely to assume the best even if you bungle your delivery in the conversation.

Show up with the right mindset

Your mindset will subconsciously impact the flow of the conversation. Take some time before the conversation to set the framing in your mind. The best is to sincerely view yourself and the other person as two humans with needs and desires that are on equal footing. You may be representing your company in the conversation, but this doesn’t make your desire more important. If you place a high value on their needs, they’ll do the same for you and the company. You’ll find collaborative and creative solutions that are far better than what happens when you impose your “more important” desires on another person.

Take the opposite scenario. If you show up blaming the other person, feeling bitterness, or feeling that your views are right and their views don’t matter much, you won’t really be able to hide it. You may deliver the message perfectly, but most people will feel the insincerity and the conversation will erode trust.

Be strategic about timing and context

Have difficult 1:1 conversations in private. This helps to establish psychological safety by reducing the involvement of ego and the need to save face in front of others.

Separately, for discussing incidents that are highly public or that involve many people (the classic example is an app-wide outage), a Blameless Post Mortem will be more productive than a private conversation.

Bringing up issues early, often, and in a supportive way can make them feel easy and casual. Small issues are often best dealt with using a quick, relaxed conversation. Bigger issues will need more time and setup.

Be mindful of the time when you approach someone for a difficult conversation. Assess how long your conversation may take, and make sure you have enough time. If a person is under exceptional stress, has an urgent need to deal with, or is distracted, wait until they have more headspace.

That said, waiting too long can be a signal of importance to the other party. If they said something hurtful last week and you bring it up this week, it signals that it’s a big deal. It appears that you may have been thinking about it for a whole week. For this reason, discussing issues sooner is generally better. Being strategic doesn’t mean waiting for the perfect time.

Having a regular standing meeting time reduces the unsettling impact of an unexpected 1:1 showing up on the calendar, and can be a more casual setting to discuss a disagreement or deliver feedback.

Make the conversation opt-in

Get consent before starting. You can ask if it’s a good time to discuss an issue and suggest that you’re also happy to discuss at another time that week if it’s better for them. This reduces the feeling that they have to say “yes” to discussing now.

Set the stage

You can create a warm, inviting space by sincerely sharing something that you admire about the person or about how they handled the situation you’re going to bring up. Acknowledge their effort. And of course, make sure that conflict isn’t the only time you praise your teammates.

You can also emphasize shared goals or context. For example, “we’ve both been in a tough situation with the shorthanded team recently” or “I know we’re both really passionate about releasing the product on time, and we’re working hard to hit that goal.” This can help you “get on the same side of the issue”¹ and create a collaborative environment for handling the disagreement. After all, neither of you want to be in a state of disagreement, and it’s a shared issue that can be best addressed by working together.

Begin with curiosity

If there’s a disagreement, you’ve likely misunderstood the situation, their intentions, or even their actions. Bring up the topic in a non-judgmental way and then listen to them with genuine curiosity. Practice active listening by paraphrasing what they’ve shared. You can tell you’re succeeding at listening rather than telling if you hear them say “that’s right” not “you’re right”.²

By listening first, you accomplish several goals:

You can correct any of your own misunderstandings. This alone may resolve the issue.
People will be more comfortable listening to you once they feel heard.
It gives them a chance to address the disagreement before you even mention it, giving them greater agency in resolving it.
Once you understand how they think about the issue, you can get your point across more effectively.

Customize

Some people are more sensitive and need only the slightest mention of a piece of feedback to fully internalize it. Others have a harder time hearing feedback or lack the context and will benefit from a more thorough discussion. Knowing your audience can make the conversation much easier.

One approach I like for delivering feedback is to progress in stages from a gentle nudge to a full explicit statement of the issue and its implications. I stop as soon as the message is received. This avoids being too heavy-handed with someone sensitive, which can be demoralizing.

People may magnify the seriousness of an issue or take it personally. If the issue isn’t a big deal, share that. Set a light-hearted tone if you can so the other person doesn’t feel that their reputation is on the line.

When more detail is necessary, Non-Violent Communication (NVC) is a great way to phrase your desires. It avoids “you” statements and accusations, sticking to what you’ve observed and how it affects you.

Other ways to keep it non-judgmental:

Avoid interpreting their behavior and making assumptions. Instead, ask them about their intentions.
Focus on what can be changed going forward and avoid dredging up past events.
Keep the conversation to the current issue rather than broadening it to other issues.

Include them in the problem solving

Ask them for their ideas on how to prevent the issue going forward. The key word there is “ask” as telling them how to change their behavior can create conflict. They’ll feel a sense of ownership if they choose how to move forward. They’ll likely come up with ideas you hadn’t thought of, which can help you learn. If it doesn’t seem like they’re coming up with strong solutions, offer suggestions. The goal isn’t only to come up with ways they can change, but also identify ways that your behavior or the greater context could be adjusted to solve the problem.

De-escalation

Sometimes a conversation will get heated. When this happens, pause and take a step back. Anger means that one or both of you isn’t really able to listen, and so the conversation won’t be productive.

A few things you can do:

If you’re triggered, slow down your reaction time to what they say and your breathing. For example, try Box Breathing.
If they’re triggered, you can mirror what they’re saying ³ to be sure they feel heard, or use the “taking the wind out of their sails approach” described by David Burns⁴.
Ask for a pause. This can be a minute of silence together in the room, a ten minute walk, or waiting until tomorrow to resume dialog.

Check in

Ask them to see if they feel heard and understood. Is anything coming up for them that they’d like to share? Confirm both of your agreed-upon action items for changing things going forward.

Also take a moment to check in with yourself. Do you feel like you were successful? Did you meet your/your company’s needs while keeping the relationship intact or even strengthening it?

Follow up

If the issues have improved down the road, recognize that, and express your appreciation to them. If not, check in with them and decide how to course-correct.

Conclusion

At the end of the day, dealing with disagreements and conflicts in a productive way can strengthen relationships and is essential to succeeding in a leadership role. Whether you’re just starting out or you’re a seasoned leader, I hope you find these tips helpful and I wish you success and continual growth on your journey.

Photos by Todd Cravens and Chris Sabor on Unsplash.

Footnotes

“getting on the same side of the issue” this excellent phrasing is from the Difficult Conversations Class run by Sara Ness ↩
One of my favorite nuggets from Never Split the Difference ↩
Another great suggestion from Sara Ness ↩
An anti-heckler technique described by David Burns ↩

Challenges in Human Healthspan Trials

Sun, 23 Jun 2019 00:00:00 +0000

No study has yet convincingly demonstrated a drug that can increase healthspan in humans. There have been randomized controlled trials (RCTs) in model organisms such as mice and c. elegans finding a wide variety of dietary, genetic, pharmacological, and exercise interventions that extend lifespan ¹, but these results have not been replicated in humans. There have been observational studies showing promising results for calorie restriction and exercise in humans, but more human RCTs are needed to be confident in the results.

The central reason for this is that a human RCT using healthspan as its primary endpoint is very difficult to run. Such an RCT faces the following challenges:

Low death rate. Only patients who reach the end of their lifespan (or healthspan) during the study contribute to the statistical power of the study. In a uniform random sample of humans, under 2% will die each year.
Slow action. If we make the reasonable assumption that the benefit of a healthspan intervention (such as calorie restriction) is roughly proportional to its duration, a study of an intervention that lasts a year will observe an effect size two orders of magnitude smaller than a study that lasts a lifetime.

Other clinical trials typically do not face these challenges. Consider a clinical trial for a blood pressure (BP) medication, for example, where the endpoint is the percentage of patients who have high BP measurements. It will typically be conducted on a population that already has sustained high BP. The ideal BP medication would theoretically be able to reduce the percentage of patients with high BP readings from 100% to 0%. In contrast, a year-long ideal lifespan intervention on a random population would only be able to reduce the death rate from about 2% to 0%.

The natural approach for remedying this issue to only include people who are likely to die soon in a healthspan trial, but this introduces new issues. Lifespan interventions are intended to help the population at large, not only very sick individuals. An intervention that is found to be effective in very sick individuals, while exciting, may have no effect or even the opposite effect in the general population. Also, it may not truly be a life extending intervention. It might merely treat a specific condition such as diabetes that is common in the sick population. The only way to demonstrate generalized healthspan extension properties of an intervention is to test it in a cohort that is close in composition to the general population.

The BP trial is also looking to find interventions that are fast-acting. An effective BP medication will take effect in weeks or months, not in years or decades. While we’d love a similar speed of action in a longevity intervention, it’s probably not realistic. The differences between a young and an aged body are present in all tissues, organs, and cells, at many different scales, and involve deep changes that have taken a lifetime to accrue. While it’s possible that fast-acting interventions exist, it’d be surprising if they were anything but a rare exception. Most interventions will probably resemble calorie restriction or exercise, which deliver incremental benefits with each additional year they are practiced. Such interventions will have a much smaller effect on healthspan during the course of a 1-2 year trial than in a lifelong trial, making the effect very hard to detect unless an enormous study is conducted.

Combining the two factors above (low death rate has a linear effect and slow action has a quadratic effect), we can derive that if we cut the length of a healthspan RCT in half but want to keep its statistical power constant, the study size must increase by roughly a factor of 8. In other words, the cost of a healthspan study skyrockets as its duration decreases. Given that drug patents last 20 years in the US, pharmaceutical companies face strong headwinds attempting to demonstrate the effectiveness of a healthspan drug to the FDA in that window and still have time to offset drug development costs by selling the drug.

A Modest Alternative: the GWAS Approach

While not suitable for an FDA trial, an alternative is to use natural experiments to build strong evidence in favor of healthspan interventions. Genetic variation is a natural experiment, as it randomly assigns different genes to different individuals for their lifetimes. It allows researchers to identify causal relationships between genes (and therefore regulatory pathways that the genes affect), and healthspan or lifespan. Genome wide association studies (GWASs) that use this method aren’t perfect (perhaps a particular gene is more common in a particular region, and healthspan is longer in that region for environmental reasons), but they’re an effective and popular methodology that has produced thousands of discoveries ².

Figure 1. GWAS methodology as shown by EBML-EBI. It identifies genetic variants that are significantly more frequent in the case group than in the control group.

That said, GWASs have largely been disappointing for explaining multi-genic conditions ³ like aging. While GWASs have identified many genetic variants that likely play a causal role in multi-genic conditions, they typically only explain a small fraction of the heritable component of the conditions. GWAS studies that focus on longevity are no exception.

Genetics have been estimated to explain between 15% and 25% of variation in longevity⁴, and that percentage more than doubles when looking at an individual’s chances of reaching age 100 once they have turned 60⁵. Given this, we can be confident that genetics plays an important role in determining longevity, but GWAS techniques only expose the tip of the iceberg. Let’s look at what we’ve learned about longevity from GWASs.

GWAS Longevity Results

Only two genes: APOE and FOXO3, have a compelling longevity story, with both showing an association with longevity in multiple independent studies. The odds of reaching 90 years of age go up by a factor of 1.2 when an individual has the beneficial variant of APOE, and 1.17 for the beneficial variant of FOXO3⁴. It’s disappointing that there aren’t more genes with stronger associations, but it’s still valuable. Let’s dive more into the two genes and the pathways they are involved in.

FOXO3

The FOXO3 gene is involved in the production of Insulin-like Growth Factor 1 (IGF-1). The IGF-1 signaling pathway (IIS) is the most evolutionarily conserved pathway of aging, shown to modulate lifespan in model organisms across a great evolutionary distance: from c. elegans to mice⁵. Dietary restriction is a well-known environmental signal shown to extend lifespan in eukaryote species, from yeast to primates. The “longevity response” to dietary restriction is regulated by several nutrient-sensing pathways: the kinase TOR, AMP kinase, sirtuins, and the IIS⁵. FOXO proteins are the most important transcriptional effectors of the IIS, which brings us to a basic understanding of the link between FOXO3 and aging. The beneficial variant of FOXO3 may create some of the same biological conditions as dietary restriction, by way of the IIS and IGF-1.

The connection between IGF-1 and longevity is complicated. IGF-1 and Growth Hormone (GH) are very similar steroid hormones. IGF-1, when given to older adults, combats muscle loss as well as cognitive decline by promoting new nerve growth⁶. IGF-1 can have beneficial effects later in life likely because the concentration of IGF-1 and GH declines as the body ages.

Figure 2: amount of GH secreted in humans declines with age. From Josh Mitteldorf.

Due to the regenerative effects of IGF-1 when given to older adults, and its apparent decline as humans age, there has been marked interest in IGF-1 for enhancing longevity. However, we also find the opposite narrative. Studies in c. elegans, drosophila, and mutagenic mouse models found that decreased levels of GH/IGF-1 delayed the onset of aging⁷. In the words of Sonntag et al.:

…two disparate concepts evolved and remain present in the literature (a) that the presence of normal levels of GH and IGF-1 accelerate aging and lack of these hormones or disruption of the signaling pathways governed by these hormones exert antiaging effects and (b) that the age-related decline in GH and IGF-1 contribute to the deterioration of physiological function and replacement of these hormones delay or reverse the aging phenotype.

Invertebrate studies consistently find that interrupting the GH/IGF-1 signaling pathway increases lifespan, and the same has been found repeatedly in dwarf, Ames mice, and GH receptor knockout mice. However, other rodent studies show inconsistent results. In many cases, no effect was found, or the increase in lifespan was only present in female mice. The effects of GH/IGF-1 deficiency appear to be species and strain specific, making it difficult to conclude what its effects may be in humans.

Direct evidence from a large meta-study in humans indicates that the relationship between GH/IGF-1 levels and longevity is U-shaped⁸. Humans with low levels tend to have cardiovascular disease, while high levels are associated with cancer mortality. Importantly, the effects of increased GH/IGF-1 appear to vary by tissue type in humans. It appears to be beneficial in neuronal and vascular tissues, but also causes insulin sensitivity and kidney disease, which decrease lifespan. This brings us to the conclusion that the FOXO3 variant likely does not extend lifespan by simply raising or lowering IGF-1 levels, but by having a more complex effect on the IIS pathway.

APOE

APOE is a gene that produces a protein (Apolipoprotein E) that is also known as APOE. It is known to play a major role in lipid homeostasis. It helps to determine the levels (and ratio) of LDL and HDL cholesterol in the body⁹. LDL and HDL cholesterol, in turn, show a strong relationship with cardiovascular disease. The ε3 variant of APOE has also been implicated in Alzheimer’s disease.

Much of what we understand about APOE’s role in aging comes from APOE knockout mice. These mice develop high cholesterol followed by atherosclerotic lesions early in their lives, and subsequently have shorter than average lifespans and healthspans. In their study, Bonomini et al.¹⁰ performed a detailed examination of APOE knockout mice from ages 6 to 20 weeks, documenting changes in morphology and biomarkers compared with normal control mice.

It’s important to note that this study examines the effects of APOE knockout, while GWAS studies in humans identified a beneficial variant of APOE. The presence or absence of the beneficial variant is distinct from complete APOE knockout, and there’s no reason they will have the same effects. However, both affect the same gene and can be expected to create their downstream effects largely via the same pathways.

In addition to previous findings that APOE knockout mice develop hypercholesterolemia and atherosclerotic plaques as they age, the authors also find fibrosis, pro-inflammatory cytokine increase, lipofuscin accumulation and a decrease of antioxidant enzymes in the kidney, liver, and heart. These conditions develop significantly faster in APOE knockout mice when compared with controls, and effect sizes are large.

Given this research, we can conclude that APOE appears to play a role in the proper function of the immune system as well as mediating factors that cause heart disease. The consequences of dysfunction in either area accumulate with age and lead to early mortality.

A Prediction

While GWAS approaches have only found longevity associations for FOXO3 and APOE, we can be reasonably confident that the pathways these genes are involved in have a direct causal connection to longevity. This is because GWASs are examining the results of natural experiments in humans, which provide higher quality evidence than observational studies on humans or RCTs on model organisms. Based on the results of human GWAS studies, we can predict that longevity drug development that focuses on the IGF-1 signaling pathway or the function of the APOE protein will likely be fruitful.

References

Fontana, Luigi, and Linda Partridge. “Promoting health and longevity through diet: from model organisms to humans.” Cell161.1 (2015): 106-118. ↩
https://www.ebi.ac.uk/training/online/course/gwas-catalog-exploring-snp-trait-associations/what-tool-name ↩
Visscher, Peter M., et al. “Five years of GWAS discovery.” The American Journal of Human Genetics 90.1 (2012): 7-24. ↩
Broer, Linda, et al. “GWAS of longevity in CHARGE consortium confirms APOE and FOXO3 candidacy.” Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences 70.1 (2014): 110-118. ↩ ↩²
Martins, Rute, Gordon J. Lithgow, and Wolfgang Link. “Long live FOXO: unraveling the role of FOXO proteins in aging and longevity.” Aging cell 15.2 (2016): 196-207. ↩ ↩² ↩³
https://joshmitteldorf.scienceblog.com/2015/09/04/hgh-and-igf-promise-and-danger/ ↩
Sonntag, William E., et al. “Diverse roles of growth hormone and insulin-like growth factor-1 in mammalian aging: progress and controversies.” Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences 67.6 (2012): 587-598. ↩
Burgers, Anne Marij G., et al. “Meta-analysis and dose-response metaregression: circulating insulin-like growth factor I (IGF-I) and mortality.” The Journal of Clinical Endocrinology & Metabolism 96.9 (2011): 2912-2920. ↩
Smith, Jonathan D. “Apolipoproteins and aging: emerging mechanisms.” Ageing research reviews 1.3 (2002): 345-365. ↩
Bonomini, Francesca, et al. “Apolipoprotein E and its role in aging and survival.” Experimental gerontology 45.2 (2010): 149-157. ↩

Crypto: The Mechanics

Mon, 15 Oct 2018 00:00:00 +0000

Cryptocurrencies center around one data structure: a distributed ledger. However, they implement it in many different ways. Here we’ll focus on Bitcoin.

Ledger: An ordered list of transactions. A ledger allows nodes on a cryptocurrency network to understand which accounts hold which coins, and detect + prevent efforts at cheating, like double-spending a coin or spending coins you never had. Usually implemented as a blockchain.

Transaction: Usually a transfer of coins from one account to another. Transactions must be digitally signed using public key cryptography, so any observer can verify that the sender approved the transaction.

Public key cryptography: Each entity generates a public key and a private key. They are effectively extremely large numbers, and they have an amazing property: data encrypted with the public key can only be decrypted by the private key, and vice versa. An entity shares the public key to everyone, and keeps the private key secret.

Digital signature (with public key cryptography): Alice wants to prove to Bob that she has signed a transaction. She shares her public key, and then adds a signature to the transaction using her private key. The signature consists of a hash of the transaction contents encrypted with her private key. Bob can then compute the same hash himself, use Alice’s public key to decrypt the signature. If he compares the hash he computed to the decrypted signature, he has proof that someone with Alice’s private key signed the transaction.

Hash:

A hash function takes an input of any length, and returns a constant-length result. SHA256 (what Bitcoin mostly uses) always produces a number 256 bits long. It is deterministic (the same input always produces the same result), one-way (given the input, the output is easy to generate, but the output tells you nothing about the input), chaotic (a one character change to the input produces an entirely different result), and collision-resistant (it is astronomically improbable that two different inputs produce the same output).

A hash is a good way to summarize a large document. If the hash of two documents is the same, you can be sure they are the same document. It’s like a fingerprint.

Blockchain:

In Bitcoin, the blockchain is a series of blocks. Each includes the hash of the last block, a Merkle Tree (we won’t get into this) of transactions in the block, and a nonce. Creating a new block requires finding a nonce (an arbitrary number) which causes the hash of the block to be extremely small. If the current mining difficulty requires 9 leading zeros on the hash, a miner must, on average, try 1 billion different nonces to produce a valid block. Since each block requires a lot of work to create, and each block contains the hash of the block before it, the entire blockchain provably took a tremendous amount of work to generate. This is called proof of work.

Proof of work: The Bitcoin protocol allows for trustless consensus. All of the nodes use the same rule for consensus: the longest blockchain (the one requiring the most work) is the correct one. If I try to create a new blockchain where I received 100 extra BTC last year, I’d have to take the chain as of one year ago, edit a block there, and then recreate all the blocks since then to have a longer chain. Unless I have access to much of the world’s computing power (like the NSA), my blockchain will always be shorter, and therefore it will be rejected.

Mining: The process of creating new blocks by trying trillions of different nonces to find a golden nonce, one that leads to a block hash lower than the difficulty. As miners add more computers (and switch to ASICs specially made to mine Bitcoin), the network adjusts the difficulty up so that a new block is mined roughly every 10 minutes.

A note about trust: While there is no trust necessary to achieve consensus on which blockchain is correct, trust in the software that most other nodes are running is still necessary. If I convinced the big miners and most users of bitcoin to run a new version of the software that has 1,000 extra BTC all owned by me, that would become the new truth. People running the old software would disagree, but they would be in the minority. This is more or less what a fork is, leading to two different versions of the currency.

Other currencies:

Ethereum: allows smart contracts, code that is run on the network and cannot be tampered with or stopped once launched. This allows Initial Coin Offerings (ICOs) that sell new coins according to predefined volume and pricing rules.
Ripple / Stellar: networks that depend on federated or centralized trust in network nodes to avoid proof of work or proof of stake consensus. They have no mining and are far more efficient.
Zcash: Bitcoin transactions are public. Zcash adds shielded transactions that are private.
FileCoin (ICO): Not yet implemented, but ICOed for >$200M. Distributed storage system.

Longevity Research

Tue, 01 Aug 2017 00:00:00 +0000

These are notes from a recent presentation I gave on the state of longevity research and potential interventions.

Context

Lifespan in animals is highly variable. Hydra are biologically immortal, glass sponges live to 10k+ years, and Ming, an ocean quahog, lived to 507.¹

The longest human lifespan was 122 years for Jeanne Calment, a French woman.

Avg. lifespan in the US is 76 for men, 81 for women.

From 1990-2014, it increased consistently at a rate of 1 year of life per 7 years.²

Biological aging is observable as an increase in mortality rate with age, and is the greatest risk factor for nearly every cause of mortality, including heart disease, cancer, diabetes, stroke, and cognitive decline.³

Healthspan: years of healthy life. Aging treatments should focus on extending healthspan.

Why do organisms age?

Most researchers agree that antagonistic pleiotropy is the leading explanation.⁴

The idea: genes that increase the odds of successful reproduction early in life may have deleterious effects later in life. Because the gene’s harmful effects do not appear until after reproduction is over, they cannot be eliminated through natural selection.⁵

Why is there hope for longevity interventions?

By the late 1990s, researchers had found that single mutations could triple the lives of of C. elegans (a small worm), and similar mutations were found for fruit flies, mice, and other organisms.⁴ This gives hope that the same could be found for humans.

However, so far, there have been repeated failures in attempts to extend human life. Resveratrol, a widely publicised molecule from red wine, still has no evidence supporting its ability to extend lifespan.⁶ Despite extensive study, multivitamins and antioxidant supplements also show no evidence of benefit and multiple vectors of harm (beta-carotene, vitamin A & E increase mortality, antioxidants may interfere with cancer treatments, increase risk of colon & lung cancer).⁷

Mechanisms of aging

Generally, aged tissues lose their capacity for homeostasis.

Oxidative stress: reactive oxygen species (ROS), primarily from respiration in mitochondria, cause an accumulation over time of damage to cellular components and DNA.⁵

Telomere shortening: Human cells can divide about 50x before the telomere becomes too short and senescence (cellular inactivity, no reproduction and altered chemistry) is triggered.⁵

Cross linking: inappropriate bonds between DNA, proteins, and other structures develop during aging, leading to dysfunction (skin wrinkles, AD plaques)⁵

Accumulation of mutations: due to ROS and other causes, mutations accumulate in tissues, leading to dysfunction and cancers5

Neuroendocrine decline: as a human ages, hormonal activity changes and becomes more erratic, leading to high blood pressure, poor sleep, and cellular dysfunction⁵

Accumulation of senescent cells: Senescent cells do not regenerate, do not fulfil their role effectively, and emit hormones that cause dysfunction globally (beyond their own tissue and organ)⁸

Depletion of stem cells: Stem cells are essential for tissue regeneration (brain, muscle, organ, etc…), but become depleted with age⁹

Epigenetics: It has been found that the expression levels of almost 100 genes changes with age. These differences may be responses to the stresses of aging or causal of aging itself.¹⁰

Immune system dysfunction/inflammation: Immune system dysfunction and inflammation increase with age. In addition to preventing infection, the immune system is important for tissue remodeling, removal of dysfunctional cells, wound healing, and regeneration. Immune dysfunction could be a key causal factor in other aging mechanisms.⁴ ¹¹

Promising interventions

Many promising interventions target the above processes.

Caloric Restriction (CR): Shown to be effective in worms, fruit flies, and mice, with mixed but promising results in primates. Researchers are investigating pathways affected by CR for a pharmacological alternative.⁴ ¹²

Young Blood: Mouse experiments have shown rejuvenation of old mice when they are given transfusions of young blood. Researchers are looking for molecular differences in young blood to find a pharmacological alternative.⁴

Metformin: First-line treatment for diabetes in use since the 50s with very few side effects. Patients with diabetes taking metformin had longer lives than a healthy control without diabetes. In clinical trials now as longevity intervention, the first trial concludes at the end of 2017.⁴ ¹³

Rapamycin Analogs: Used as immune system suppressant during organ transplants. Shown to rejuvenate elderly immune systems by Novartis in 2014. Elderly patients, who respond less to flu vaccines, had an improved response after 2 weeks of taking a rapamycin analog. The rapamycin pathway (mTOR) tweaks immune systems and suppresses deleterious senescent cell secretions. It also extends mouse lifespans.⁴ ¹²

Senolytics: Drugs that eliminate senescent cells and increase lifespan in rodent models.⁸ ¹²

What can I do now to extend my healthspan?

Exercise: Shown to reduce obesity, risk of heart disease, stroke, diabetes, neurodegeneration, cancer, and all-causes mortality.¹⁴ ¹⁵ ¹⁶

Nutrition: Increasing consumption of whole grains, vegetables, fruits, nuts, and fish is associated with reduced all-cause mortality, while increased red meat and processed meat is associated with increased mortality.¹⁷ However, these results seem less well established than exercise.

Sunscreen: Daily use causes 24% reduction in skin aging in RCT,¹⁸ as well as reducing risk of some skin cancers.¹⁹

References

https://en.wikipedia.org/wiki/List_of_longest-living_organisms ↩
https://www.google.com/search?q=us+lifespan ↩
Kaeberlein, Matt, Peter S. Rabinovitch, and George M. Martin. “Healthy aging: The ultimate preventative medicine.” Science 350.6265 (2015): 1191-1193. ↩
http://cen.gext.acs.org/articles/95/i10/hit-snooze-button-aging.html ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷
De, Abhijit, and Chandan Ghosh. “Basics of aging theories and disease related aging-an overview.” PharmaTutor 5.2 (2017): 16-23. ↩ ↩² ↩³ ↩⁴ ↩⁵
https://en.wikipedia.org/wiki/Resveratrol ↩
https://en.wikipedia.org/wiki/Antioxidant ↩
Childs, Bennett G., et al. “Cellular senescence in aging and age-related disease: from mechanisms to therapy.” Nature medicine 21.12 (2015): 1424-1435. ↩ ↩²
Oh, Juhyun, Yang David Lee, and Amy J. Wagers. “Stem cell aging: mechanisms, regulators and therapeutic opportunities.” Nature medicine 20.8 (2014): 870-880. ↩
De Magalhães, João Pedro, João Curado, and George M. Church. “Meta-analysis of age-related gene expression profiles identifies common signatures of aging.” Bioinformatics 25.7 (2009): 875-881. ↩
Brown, Bryan N., Brian M. Sicari, and Stephen F. Badylak. “Rethinking regenerative medicine: a macrophage-centered approach.” Frontiers in immunology 5 (2014). ↩
Kirkland, James L. “Translating Advances from the Basic Biology of Aging into Clinical Application.” Experimental gerontology 48.1 (2013): 1–5. PMC. Web. 8 July 2017. ↩ ↩² ↩³
https://clinicaltrials.gov/ct2/show/NCT02432287 ↩
Ahlskog, J. Eric et al. “Physical Exercise as a Preventive or Disease-Modifying Treatment of Dementia and Brain Aging.” Mayo Clinic Proceedings 86.9 (2011): 876–884. PMC. Web. 8 July 2017. ↩
https://en.wikipedia.org/wiki/Physical_exercise ↩
http://www.cochrane.org/CD002968/ENDOC_exercise-for-type-2-diabetes-mellitus ↩
Schwingshackl, Lukas, et al. “Food groups and risk of all-cause mortality: a systematic review and meta-analysis of prospective studies.” The American Journal of Clinical Nutrition 105.6 (2017): 1462-1473. ↩
Hughes MCB, Williams GM, Baker P, Green AC. Sunscreen and Prevention of Skin Aging: A Randomized Trial. Ann Intern Med. 2013;158:781-790. doi: 10.7326/0003-4819-158-11-201306040-00002 ↩
https://en.wikipedia.org/wiki/Sunscreen ↩

Nuclear Energy: Costs & Benefits

Sat, 01 Jul 2017 00:00:00 +0000

These are notes from a recent presentation I gave on nuclear energy and the considerations surrounding it.

Nuclear Construction

~450 reactors world wide

~70 under construction, mostly in China, where coal power air pollution is an urgent issue

~5 under construction in the US, the first permits issued in 34 years

Coal: Historically, coal mining has been a very dangerous activity and the list of historical coal mining disasters is a long one. In the US alone, more than 100,000 coal miners were killed in accidents over the past century,¹ with more than 3,200 dying in 1907 alone.² In the decades following this peak, an annual death toll of 1,500 miner fatalities occurred every year in the US until approximately the 1970s.³ Death rates per year between 1990 and 2012 have continued to decline, with in the US below 100 coal mining fatalities per annum occurring each year during this period.⁴

Hydro: Dominated by one disaster in which Typhoon Nina in 1975 washed out the Shimantan Dam (China) and 171,000 people perished.

Oil: The rupture of the NNPC petroleum pipeline in 1998 and the resulting explosion cause 1,082 deaths.

Nuclear: The 1986 steam explosion at the Chernobyl nuclear reactor in the Ukraine caused approximately 4000 deaths in total, to eventually result in the decades ahead due to the radio-isotope pollution released.

Carbon Emissions Fatalities: By far the greatest energy fatalities that result from energy generation by humanity, is the creation of air pollution. The most lethal of which, particulate matter, is primarily generated from the burning of fossil fuels and biomass and is estimated to cause 2.1 million deaths annually.⁵ ⁶

Nuclear emissions (full lifecycle): 16g CO2 eq/kWh, similar to wind & solar

Coal and natural gas: 820 and 490 g CO2 eq/kWh

Nuclear Storage

There are some 65,000 tons of nuclear waste now in temporary storage throughout the U.S.⁷ Since 1987, Yucca Mountain, in Nevada, had been the proposed site for the Yucca Mountain nuclear waste repository, but the project was shelved in 2009 following years of controversy and legal wrangling.⁸ ⁹ An alternative plan has not been proffered.¹⁰

Without a long-term solution to store nuclear waste, a nuclear renaissance in the U.S. remains unlikely. Nine states have “explicit moratoria on new nuclear power until a storage solution emerges”.¹¹ ¹²

There is an “international consensus on the advisability of storing nuclear waste in deep underground repositories”,¹³ but no country in the world has yet opened such a site.

Cost Competitiveness

New Reactor Designs

Passive safety systems
Reduced cost of construction & operation (but not yet clear exactly how much)

Controversy

Observers attribute the nuclear controversy to the impossibility of generating a shared perception between social actors over the use of this technology¹⁴ as well as systemic mismatches between expectations and experience.

Arguments in favor:

Renewable energy and energy efficiency may not solve the energy and climate crises
Nuclear fuel is virtually unlimited and has extremely high specific energy
New technology may be able to safely dispose of nuclear waste
Nuclear power is claimed to be the safest energy option
Advanced nuclear power may strengthen global security
Nuclear power’s true costs are claimed to be lower than either fossil fuels or renewables
Nuclear power may lead the “clean energy” revolution

Arguments against (from Lowe):¹⁵

It may not be a fast enough response to climate change
It is claimed to be too expensive
The need for baseload electricity may be exaggerated
The problem of waste may still remain unresolved
It may increase the risk of nuclear war
There are claimed to be major safety concerns
There are claimed to be better alternatives

Decommissioning

After a cooling-off period that may last as long as a century, reactors must be dismantled and cut into small pieces to be packed in containers for final disposal. The total energy required for decommissioning can be as much as 50% more than the energy needed for the original construction. In the U.S. there are 13 reactors that have permanently shut down and are in some phase of decommissioning, and none of them have completed the process.

Federal Subsidy

There is a federal insurance subsidy for incidents beyond cost of $10bn in the US.

Radioactivity from coal

Coal plants release more radioactivity into the environment than nuclear plants, through the release of thorium and uranium in coal ash.¹⁶

Long-term effects of working at or living near nuclear power plants

Studies conducted by non-profit, neutral agencies have found no compelling evidence of correlation between nuclear power and risk of cancer.¹⁷

Tom Quisel

The Future of Agriculture

Crafting the job candidate experience

1. Move quickly

2. … and also be thorough

3. Paint a clear picture of the interview process

4. Form a real human connection

5. Be transparent about your company

6. Have a diverse interview panel

7. Keep it comfy

A great experience

Navigating Disagreements as a Leader

It’s easier if you have a good relationship

Show up with the right mindset

Be strategic about timing and context

Make the conversation opt-in

Set the stage

Begin with curiosity

Customize

Share your desires non-judgmentally

Include them in the problem solving

De-escalation

Check in

Follow up

Conclusion

Footnotes

Challenges in Human Healthspan Trials

A Modest Alternative: the GWAS Approach

GWAS Longevity Results

FOXO3

APOE

A Prediction

References

Crypto: The Mechanics

Other currencies:

Longevity Research

Context

Why do organisms age?

Why is there hope for longevity interventions?

Mechanisms of aging

Promising interventions

What can I do now to extend my healthspan?

References

Nuclear Energy: Costs & Benefits

Nuclear Construction

Energy-related Fatalities

Nuclear Storage

Cost Competitiveness

New Reactor Designs

Controversy

Decommissioning

Federal Subsidy

Radioactivity from coal

Long-term effects of working at or living near nuclear power plants

References