Why it matters: The cannabis industry spent thirty years breeding for one number. Now its most lucrative potential customers are the ones that number scares off.

The weed of yesteryear was a different plant. Flower seized by the DEA in 1995 averaged about 4% THC, according to the University of Mississippi’s potency monitoring program. By 2022 it was over 16%. Legal flower in Colorado now averages around 21%, and concentrates run about 71%. One joint today is roughly five from your days in college.

Nobody asked for this. Breeders selected for potency because dispensaries sort their menus by THC percentage, and shoppers treated the number as a quality score. Mild came to mean cheap.

The customers are already here. Cannabis use among adults 55 to 65 has surged, and the industry has noticed. Forbes says low-dose products are becoming the industry’s growth engine. Forty-two percent of edible buyers now prefer 10 milligrams or less. Dispensaries are adding low-dose and 1:1 CBD options.

The catch: It’s all gummies, seltzers and tinctures. Nobody has built a flower brand around going easy. The people who actually want to roll one up and share it on the porch are stuck choosing between rocket fuel and nothing.

The idea: Grandpa’s Homegrown. A premium flower brand at 6 to 8% THC, with a real CBD ratio, sold on flavor and feel instead of a number.

• Sell the session, not the strength. Beer figured this out. A 4% session IPA isn’t a worse beer than a 12% barleywine. It’s a different occasion. Wine figured it out too: nobody brags that their Pinot is stronger than Everclear.

• Lean into the nostalgia. Brown paper, a hand-drawn grandpa on the label, strain names like Porch Swing and Dylan Went Electric. It’s the weed you remember, grown by people who know what they’re doing now.

• Balance the CBD. The old landrace strains carried a meaningful CBD ratio, which is part of why they felt gentler. Modern breeding bred it out, but we need it back.

• Make it shareable. Pre-rolls sized for two or three people over an hour, not one person over fifteen minutes.

Why nobody’s done it: The whole supply chain points the other way. Growers get paid per percentage point. Budtenders upsell on potency. And you’d be the only brand on the shelf advertising a lower number. That takes nerve, and a marketing budget.

Disclosure: I have no financial interest in Grandpa’s Homegrown, but I do have an editorial one. Mild weed might inspire five Substack posts a month out of a dad in his 40s instead of two.

The bottom line: The weed industry won the potency race and lost a generation of customers who just wanted to feel a little nicer.

Curaleaf or Trulieve, please steal this idea.

Midway tried. Twice. In the mid-’90s, the studio behind NBA Jam built Power-Up Baseball, put a handful of cabinets in Chicago arcades, and pulled them within a couple of weeks. The problem was length. “You can’t predict the gameplay time,” its designer said. Midway’s later MLB Slugfest had an on-fire mode, but it kept seven innings and the base paths.

Both kept the slowest parts of baseball.

The idea: MLB Slam. Three-on-three backyard Wiffle ball with real big-leaguers. No base running — just pitching and hitting.

• Backyard rules. Ghost runners instead of base running. Over the fence is gone. Three innings, done before anyone calls you in for dinner.

• Three hits in a row and you’re on fire. Flaming bat, bigger sweet spot, more pop. You stay hot until you make an out.

• Every home run earns a bat flip. The higher and the more rotations, the bigger the style points. Baseball spent a century calling that bad manners. Then MLB’s own 2018 playoff ad had Ken Griffey Jr. telling everyone to let the kids play.

Don’t believe me? Play it. I used Claude to draft this post, and while I was tweaking the text, it just built this: Backyard Slam. Arcade graphics, a neighborhood crowd, flaming bats, and a bat-flip meter. It works on your phone.

A window-smashing homer, a 95-foot bat flip, and a pitcher on fire. (Play Backyard Slam)

The game was the easy part. The names on the lineup card are not.

Tater, Flapjack, and Puddlejump, standing in for real big leaguers.

The catch: Everything that makes it MLB Slam belongs to someone else.

• Two locks, two keys. MLB controls the team names, logos, and uniforms. The players’ union controls every active player’s name, image, and likeness. Backyard Baseball needed both licenses to put 31 big-leaguers in a kids’ game.

• Rights don’t last. When MLB Slugfest was remastered in 2025, Barry Bonds, Frank Thomas, and Ken Griffey Jr got swapped for generic stand-ins. Even NBA Jam never had Michael Jordan, who controlled his own likeness.

• Labor is the wild card. The players’ contract expires at 11:59 p.m. on December 1, and owners are expected to lock them out the next day. The last time that happened, MLB scrubbed current players’ photos from its own website.

• Even the ball is trademarked. “Wiffle” belongs to The Wiffle Ball, Inc. of Shelton, Connecticut.

The bottom line: In 1993, NBA Jam took a team of engineers and artists. Today, a playable demo shows up in the time it takes to draft a blog post. The code is cheap now. The rights never will be.

MLB Players Inc., please steal this idea. I’m not building the real thing. I just want to play it.

This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.

Why it matters: Touch came from the iPhone. The interface the Mac actually needs is trapped inside a $3,499 headset that almost nobody wears.

I’ve seen the future. The first time I tried on the Apple Vision Pro, I saw the future of computing. You look at something and it’s selected. You pinch your fingers and it opens. It’s the Minority Report interface we’ve been promised since 2002, except this one actually works. You’re not using the computer. You’re literally in the computer.

The catch: You look like a complete dork.

Exhibit A. (Photo: Bram Van Oost / Unsplash)

Exhibit B: You pretty much never see anyone using one in public. Apple has sold roughly 600,000 since launch, returns have reportedly run unusually high, and the company has all but stopped work on it.

The magic was never the goggles. It was the eyes and the hands.

The idea: Bring Vision Pro’s eye tracking and hand-gesture recognition to the MacBook Pro through an expanded notch.

Apple just went the other way, shrinking the notch into a Dynamic Island. I’d make it bigger and pack it with the sensors that let Vision Pro know where you’re looking and what your hands are doing. Look at a window to focus it. Pinch in the air above the keyboard to click. Flick to scroll.

It might be even better than a touchscreen:

• No gorilla arm. Steve Jobs said it best back in 2010: “Touch surfaces don’t want to be vertical.” Reaching up to poke a screen gets old fast. With gestures, your hands barely leave the keyboard.

• No smudges on a $2,899 OLED display.

• Your eyes are already on the screen. Let them do the pointing.

It’s less of a moonshot than it sounds. Apple already ships eye tracking on iPhone and iPad using nothing but the selfie camera.

The pièce de résistance: Turn the trackpad into an e-ink surface that takes a stylus. Sketch a diagram, sign a contract, or mark up a draft without ever lifting your hands from the laptop. (Funny enough, the new touchscreen doesn’t even work with Apple Pencil.)

Lenovo has flirted with pieces of this. Its Yoga Book C930 had an E Ink panel you could write on, and its Yoga Pro Aura Edition laptops let you sketch on the trackpad. Nobody has put it all together.

The bottom line: Eyes to point. Hands to gesture. A pen to create. No goggles required.

Apple, please steal this idea.

The Situation

Frontier research now runs on compute, and universities can’t afford it. Studying how AI systems behave, fail, and interact in the world takes clusters that cost billions. Not even Harvard can buy a seat at that table.

Case in point, Harvard, with an endowment of nearly $68 billion, is weighing about $25 million for a shared GPU cluster. That would roughly double its AI computing capacity, but it’s less than 0.04% of the endowment and a fraction of a single frontier training run. If Harvard can’t keep up, no university can.

So the most important research of the decade is happening almost entirely inside a handful of private labs. That’s a problem for science, which needs independent checks. And it’s a problem for students, who learn the frontier by working on it.

We’ve solved this kind of problem before. The Morrill Act of 1862 gave each state 30,000 acres of federal land per member of Congress. States sold or used it to found colleges that taught agriculture, engineering, and practical science to ordinary Americans. Cornell, MIT, Penn State, Purdue, and the University of California all trace roots to it.

It may be the best bet on public knowledge this country ever made.

The Idea

A 21st Century Digital Land Grant. Instead of acres, we grant compute. Here’s the sketch:

• Who gives: any data center above a size threshold, say 100 MW, so small operators are untouched.

• What they give: 10% of capacity, delivered as real GPU hours on their own hardware. No cash buyouts that let the frontier stay private.

• Who gets it: R1 and R2 research universities, which means public flagships and regional schools, not just the Ivies.

• How it’s shared: a neutral clearinghouse like the National Science Foundation’s existing AI research resource, scaled up to allocate hours by peer review.

• What it buys: open research on safety, alignment, science, and medicine, published in public.

The beauty is that nothing new gets built. The capacity is already coming online. We’re just reserving a public lane on the highway.

Why the builders might say yes

Morrill gave away public land, but this asks something of private builders.

Then again, data centers aren’t purely private. They run on public power grids, public water, public land, and public tax breaks. Towns are starting to push back on all four. A share-back to the public is the price of a social license to build.

Yes, it’s effectively a tax on the hyperscalers. But it’s a tax that pays for the research, talent pipeline, and independent scrutiny the frontier needs anyway. Today’s grad student on a granted cluster is tomorrow’s lead researcher at the lab that donated the compute.

The hard part

But if one state requires it, won’t builders just move next door? Probably. A state-by-state set-aside invites a race to the bottom.

States are already trying the other route by buying compute outright. New York’s Empire AI pools nine universities, including Columbia, Cornell, NYU, CUNY, and SUNY, around shared GPU clusters. The NSF’s National AI Research Resource pilot has backed 500+ projects and is becoming permanent. Both are great. Both are a rounding error next to what the labs spend.

So the real version has to be national. Here are three possible paths:

1. Executive order. Tie the set-aside to what Washington already controls: federal contracts, federal land, and fast-track permitting and grid hookups. Likely the next President, not this one.

2. Carrots before sticks. States trade faster permits and power deals for a voluntary grant. Builders flock to the states that offer it.

3. Congress. The most durable solution is an act from Congress, but today that seems like the least likely to happen.

One reason for hope: this is a rare idea with friends on both flanks. Bernie Sanders and Steve Bannon both say Big Tech should give something back for what it takes from communities. When those two agree, it’s worth getting curious.

Please steal this idea

The Morrill Act didn’t just build the higher ed system we know today, it decided who got to take part in the economy of the last century. If you’re a governor, a senator, a university president, or a hyperscaler looking for a better story than “we drove up utility prices,” this one’s yours. Take it. Improve it. Put your name on it.

Please, someone make the 21st Century Digital Land Grant happen.

What if every school that took federal student aid had to buy insurance on its students’ debt, and the policy only paid out if the school actually did its job?

The problem we keep paying for

Right now, when a student borrows, enrolls, and leaves without a degree, the school keeps the tuition, the student keeps the debt, and taxpayers eat the default.

Everyone pays except the one party that could have changed the outcome.

37.6 million working-age Americans have some college and no credential. And borrowers who don’t finish are more than twice as likely to default as those who do (59% vs. 23% in Pew’s survey).

Debt without a degree is the worst deal in higher ed, and the school bears none of the downside.

The idea: insure the diploma

Every school that takes federal aid buys a policy on its students’ loans. If a student borrows and doesn’t graduate, the policy pays down the debt.

Suddenly the school has skin in the game. Not through a fine or a federal mandate, but through a premium it pays every year.

Graduate more students, and pay less. Lose them, and you’ll pay more.

Students chip in a sliver too, so everyone has a stake in finishing.

The analogies

Car insurance. Your rates depend on your driving record. After a crash, your insurer fixes your car first and sorts out fault later. If the other driver caused it, your insurer collects from them. If you caused it, you pay more. Here, the loan is the car, the pool is the insurer, and the school is the other driver.

GAP insurance. If your car gets totaled while you still owe more than it’s worth, GAP covers the difference. Its cost is often rolled right into the loan. You’re not stuck making payments on a car you can’t drive. Debt without a degree is the same trap: payments on a diploma that never showed up.

The FDIC. Banks pay for deposit insurance at rates priced by risk, set partly by the CAMELS ratings their examiners hand out. Safer bank, cheaper premium. This is basically the FDIC for diplomas.

Bond ratings. Moody’s already rates colleges. It just rates whether they’ll repay their bondholders, not whether they’ll graduate their students.

The catch: you’re only covered if you did the work

The policy is compliance-gated. Follow the best practices and the pool covers the loss. Fall short, and the student still gets covered, but the school pays the pool back.

That kills the moral hazard. A school can’t run a sloppy operation and dump the cost on the insurer. Medicine has a standard of care. Higher ed doesn’t. This would create one.

What counts as best practice? Mostly things we already know work:

• Advising ratios that let someone actually know each student

• Early-alert systems that catch students before they slip away

• Proactive advising and just-in-time interventions

• Financial aid and emergency aid that show up on time

• Honest, public outcomes data

Don’t build a new regulator. Use the one we have.

Roll the audit into accreditation. Accreditors already do deep institutional reviews on a regular cycle. They already have standards. They just don’t have teeth.

Today accreditation is basically binary. You’re in or you’re out, and pulling it is a nuclear option nobody wants to use. Tie insurability to the review, and every finding suddenly has a price.

It also fixes the coziness problem. Accreditors get accused of being too close to the schools they review. Add an insurer with real money on the line, and someone in the room finally has a reason not to go soft.

A credit score for colleges

You have a credit score for paying your credit card bill. Schools should have one for graduating their students.

The rating moves, and it sets the premium. A school sliding from AA to BBB feels it in the budget years before it’s anywhere near losing accreditation. That’s a gradient instead of a cliff. An early warning instead of a funeral.

And families get to see it before they borrow a dime.

One rule for the score: measure value added, not raw graduation rates. Judge each school by its outcomes relative to the students it actually enrolls. Otherwise the cheapest move is to stop admitting first-gen and low-income students, which is the exact opposite of the point.

How the money flows

Everyone pays in. Whoever dropped the ball pays out.

It works like a fender bender: the student gets covered first, then fault gets sorted out.

1. Everyone pays in. The school pays a yearly premium set by its rating. The student pays a sliver of what they borrow, rolled into the loan like GAP coverage.

2. Nobody has to file a claim. If a student goes a full year with no degree and no enrollment anywhere, the claim triggers on its own. The loan servicer files it, using enrollment data schools already report.

3. The student gets covered first. The pool pays down the loan, minus a deductible. A student who showed up keeps a small one. A student who ghosted, skipping class and ignoring every outreach, keeps some or all of the debt. Real hardship gets an exception.

4. Then the pool settles up with the school. It checks the school’s latest accreditation review. If the school met the standard of care, the pool eats the loss. If it didn’t, the school pays the pool back. That’s the chargeback. And a student who showed up gets their deductible back, just like when your insurer collects from the other driver.

💸 The bottom line: Students who try get covered first. Schools pay only when they didn’t do their part.

Why now

Washington already wants schools to have skin in the game. It just reached for a blunt tool.

In 2025, the House version of the big reconciliation bill would have made colleges repay part of the federal loans their former students didn’t. The Senate swapped it out for an earnings test instead.

Penalties are a political fight. Insurance is a market. Premiums, ratings, and audits are things every college CFO already understands.

What I’m NOT prescribing

• Who underwrites it: private insurers, a federal pool, or a mutual owned by the schools themselves

• Whether it’s required for every school taking federal aid or starts as a voluntary pilot

• The premium formula

Those answers should come from people who know actuarial math a lot better than I do.

Open questions

• Who counts as “didn’t graduate”? How do we handle stop-outs who come back? Or someone who leaves for a great job?

• How to stop schools from gaming the audit, checking every box while still failing students?

• How do you keep schools from lowering the bar just to push people across the finish line?

• What happens when a school closes mid-policy?

• Could the same model cover other forms of education, such as bootcamps, certificates, and income-share agreements?

Possible names

• Finish Line

• The Diploma Guarantee

• Completion Coverage

• Commencement Insurance

Please steal this idea. And if you do, tell me how it goes.

~Drew

Sources

• Some College, No Credential: A 2025 Snapshot, National Student Clearinghouse Research Center, June 2025

• Borrowers With Certain Educational Experiences Appear More Likely to Default, Pew Charitable Trusts, January 2024

• Risk-Based Assessments, FDIC

• Holding Colleges Accountable: From Graduation to Gainful Employment, American Enterprise Institute, July 2025

• Senate Reconciliation Bill is a Better Version of a Bad Idea, American Council on Education, June 2025

• What is Guaranteed Asset Protection (GAP) insurance?, Consumer Financial Protection Bureau

Because organisms with modified genetic sequences have been viewed as machines their production has been protected under patent law. Therefore, like other machines, the only way to make a profit from their production has been through control of the means and manner of production. It would make no sense to go to all the trouble of developing a new mousetrap if the buyer could simply reproduce it for their own use or sale to others. Therefore companies in the business, like Monsanto, have gone to great lengths both physically and contractually to make sure that any farmer who buys their transgenic seed not be able to use the crop from that seed to plant more crops. Under this system, farmers must buy seed every year from Monsanto to get the advantage of using their seed and Monsanto must maintain huge acreages of land solely for the purpose of producing seed for the next year. The investment and the infrastructure on Monsanto’s part are gigantic and they must take the risk that their product will be in demand the following year. Also the farmer takes a risk, as do his neighbors, that if there is cross pollination with other crops, Monsanto may make a charge of patent infringement on anyone downwind. 

However, consider an alternative approach. A genetic sequence from a synthetic biology viewpoint is only a sequence of the letters that make up the base pairs in all life on earth. Just as the 26 letters in the Roman alphabet can be combined and rearranged to make a novel which is protected by copyright, the four letters of the DNA alphabet can be combined and rearranged to make a gene. Viewed this way, genes should be protected under copyright law rather than patent law. The advantages to producers of new genes, buyers of these organisms, and society as a whole would be enormous. 

For the producer of the modified organism, the advantage would be that merely the creation of the new seed or fish or tree is all that is required to establish a right to be reimbursed. (Note: In case there might be any concern about someone claiming independent development, it is possible not only to insert new active gene sequences in an organism but also coded inactive sequences that would establish ownership, much as a copyright © in a book.) It would no longer be necessary for the creator of the new organism to “publish” every copy. Just like with print, anyone who makes a copy would owe a fee so the originator would encourage people to make copies. The buyers of these organisms could decide how many to sell as product (at which time the fee would be collected) and how many to use to make more copies. They might even sell their produce to others so they could make more copies (for which they would pay a fee to the originator). And society would benefit because distribution of the best products for each area would flourish and the cost to develop new organisms would be low enough that entry of competitors would be enhanced, thus keeping costs to consumers in balance.

If someone can work with companies like Monsanto to understand the advantages of the patent system, including a much longer period of protection (80 years vs 17), much lower capital investment in infrastructure, and a much larger market for additional products perhaps they might champion this approach instead of opposing it. Of course in order for this to work a system for identifying products for sale at a major middleman level (e.g. Archer Daniels Midland) would need to be implemented and a standard for testing and collection of fees would need to be developed at a national and probably an international level.  However since most crops for the general market flow through a relatively small number of middle men, this choke point should easily catch the vast majority of products on their way to market. 

It is also possible that the profitability of some of the small developers for whom the copyright system would have wide appeal could overtake the large companies, just as Microsoft overtook IBM, and implement the system without their consent.

— Jim Jarmusch

The idea is to split combo offers that are offered in restaurants and other retailers among strangers looking to get the best deal possible.

Let’s say a restaurant offers “a free drink with an order of fries,” but you only want fries.  Now, we can use technology to find someone else interested in buying a coke to help you both get the best deal possible.  

An app could enable consumers to group-buy products at a discount and bring new customers to businesses.  This could work for so many other business types: sneaker deals, clothing, glasses, you name it.

The stores could post their combo offers on the website or consumers could do so for them.  When customers want to buy something they just go on the app, pick an item, and find a purchasing partner.

The app could manage payments or use a third party, like Venmo, to transact.  It could also afford the opportunity to meet new people with similar commercial interests.

This would likely catch on among young people (with little money) before spreading to everyone else.  

Toker is a location-based app to find marijuana-smoking partners nearby.  As a toker, you can bring the necessary means or partake in someone else’s herbal generosity.  Since pre-colonial times, the peace pipe has brought people together.  It’s time to bring back the practice that transcends culture, age, and economics… at least for people over 21 in Colorado and Washington. We could also measure contribution vs. consumption to identify users like this:

With The Dating Ring, you apply to join. Pass the first bar and you’ll meet in-person with a Dating Ring matchmaker for five minutes. Users pay $25 for the initial matchmaker meeting and $20 per date. 

Sounds a little like PSTI's Cyrano for online dating… all they have to do next is tell you how to woo her.  

Scholars have differed about the correct interpretation of these dark words, which may mean no more than that the fox, for all his cunning, is defeated by the hedgehog’s one defense. But, taken figuratively, the words can be made to yield a sense in which they mark one of the deepest differences which divide writers and thinkers, and, it may be, human beings in general.

For there exists a great chasm between those, on one side, who relate everything to a single central vision, one system less or more coherent or articulate, in terms of which they understand, think and feel-a single, universal, organizing principle in terms of which alone all that they are and say has significance-and, on the other side, those who pursue many ends, often unrelated and even contradictory, connected, if at all, only in some de facto way, for some psychological or physiological cause, related by no moral or aesthetic principle; these last lead lives, perform acts, and entertain ideas that are centrifugal rather than centripetal, their thought is scattered or diffused, moving on many levels, seizing upon the essence of a vast variety of experiences and objects for what they are in themselves, without consciously or unconsciously, seeking to fit them into, or exclude them from, any one unchanging, all-embracing, sometimes self-contradictory and incomplete, at times fanatical, unitary inner vision.

The first kind of intellectual and artistic personality belongs to the hedgehogs, the second to the foxes; and without insisting on a rigid classification, we may, without too much fear of contradiction, say that, in this sense, Dante belongs to the first category, Shakespeare to the second; Plato, Lucretius, Pascal, Hegel, Dostoevsky, Nietzsche, Ibsen, Proust are, in varying degrees, hedgehogs; Herodotus, Aristotle, Montaigne, Erasmus, Molière, Goethe, Pushkin, Balzak, Joyce are foxes.

~ Isaiah Berlin

It’s probably not possible to make this app on iOS (unless it’s a jailbroken phone with extra api calls), but I think it’s doable on Android.

Here is a video clip from the show explaining Charlie’s app in very fast dialogue:

On January 25th, Bill Barnwell reflected on Tom Brady’s career in reverse chronological order: “He’s the guy who beats up on weak links and never shows up when his team really needs him. He’s got various maladies, he chokes when his team is a huge favorite at home, he can’t produce a drive to kill off a game.”

Barnwell notes that many critics had similar remarks about Peyton Manning earlier in his career, but that all changed in 2007 when Peyton won the Super Bowl. Now, the question is whether the tides area again shifting for these two quarterbacks.

While Peyton is breaking passing records with Wes Welker’s help, the Patriots are winning old school. The inverse relationship of Tom and Peyton’s careers is so surreal it brings to mind pop culture Disney classic “Freaky Friday.” Like Lindsay Lohan and Jamie Lee Curtis in the recent version (or Barbara Harris and Jodie Foster in the original), it’s as if Tom and Peyton have swapped pigskin personalities.

If you take for granted this zero-sum game of football karma, the obvious question is “when exactly did these two exchange quarterback qualities?”

It would appear the first shift occurred at halftime of the 2006 AFC Championship game in Indianapolis. After a thumping in the first half, Peyton turned the tables in miraculous fashion in the second half on the way to winning the ‘07 Super Bowl. Meanwhile, Brady and Belichick went home vowing never to play another road playoff game and assembled the world’s most outrageous offense, powered by Wes Welker and Randy Moss.

Brady’s injury and losses to the “other” Manning are trials too painful to recount. Was this change in karma penance for spying, offensive hubris, or bagging a supermodel wife? We don’t know, but it does appear circumstances are turning once again.

Peyton has the record-setting offense, made possible by Brady’s erstwhile Jonah at wide receiver, and the Patriots have rushed their way to the AFC Championship against injurious odds. This latest denouement, you may be wondering, appears to have been halftime of this season’s matchup in Foxboro.

Also the tale of two halves, that Patriots-Broncos game saw the football gods change allegiances as if by the flip of a coin. He was once the improbable David against the Rams, then Goliath against the Giants, now hopefully we have the Tom of old against Peyton’s Broncos. Do you doubt this theory? We’ll see what happens this Sunday afternoon.

The Idea: Learn it.  Do it.  Teach it.  Coding requires intelligence, creativity, and experience. Like the trade of blacksmith or art of kung fu, it helps to have some instruction, lots of practice, and an opportunity to share what you’ve learned.  StackOverflow is an amazing site for coding Q&A, but we don’t yet have easy access to personalized coding mentorship.

The Product: A website that connects aspiring web developers with more experienced coders for remote pair programming for an hourly instructor’s fee.  After completing a short survey about languages and knowledge, you would connect with someone online.  The experienced coder would determine the project (ideally an open source one) and her understudy would pay to watch and help.  Imagine having an apprentice pay you to help you with your work!  If it’s a bad match of skills, no worries.  There is another potential pair just a few clicks away.

The Market: People are willing to pay vast sums to learn to code and individual instruction is the fastest path.  There are plenty of open source projects that could use contributors at all levels.  Plus, mentors could easily identify talented students for job opportunities or to become mentors themselves.

— Osho