When context changes, so do the prospects of these ideas.
Youtube wasn't the first video streaming service but it was one of the first for the DSL era when people could watch video without lengthy waits.
AI companies repeatedly failed until enough things, specifically data and compute were at enough scale to deliver.
Advancements in battery technology made electric cars practical bucking the trend of decades of failed EV car companies.
So many things - contactless payment, touchscreens, even LCD panels, these were lousy and impractical for decades.
Attempts at mass adoption of handheld computers, now called smartphones, started in the 1980s. Without high speed mobile networks, high density color LCD screens, reliable geolocation, these things were necessary to make the handheld pocket computer something that everybody has.
Even online grocery delivery services, now common place, had its start in the catastrophic collapse of WebVan in the 1990s. Cell phones, the gig economy, mature e-payments, these were all needed.
You always need to look for the context change and how that can untar some tarpits.
On the subject of context I wrote a short post back in 2018 [0].
Smart people spend time on this problem/solution.
Solutions appear but fall short of expectations.
The technology or more commonly that application of it is stigmatised.
Sometimes the whole field becomes tainted.
The problem/solution complex is declared a âdead endâ or âfalse dawnâ.
Interest cools. Nobody invests for a while. The wreckage of the previous cycle rusts away. The craters erode. This takes 20-30 years.
During this period, some very small companies, academics, and individuals continue to guard the flame, but lack funding or new talent to advance.
Go to step 1, invent new buzzwords/framing and repeat.
Ignore much of what was learned during the previous cycle.
[1] https://blog.eutopian.io/the-next-big-thing-go-back-to-the-f...
I remember the excitement around VR back in the late 80s. These new polhemous motion tracking devices and LED microdisplays were going to change the world! Except the technology was expensive and ultimately it kinda sucked. It was barely used outside academia, interest died off gradually and eventually it was tacitly acknowledged to be going nowhere.
Then 30 years on Oculus was founded and everyone who'd never used one of the old VR systems was super excited. To be fair, the technology was a step better - much cheaper and more accessible, low motion input latencies, better resolutions. But ultimately it's still not really quite good enough and it seems that the hard reality is it's not going to make its way into mainstream consumer everyday use this time either.
I can't wait for round 3 in 2040 or so.
> But ultimately it's still not really quite good enough
I'm not sure what use cases you've tried it, but I'm "playing" a bunch of flight simulators, and after getting used using a HP Reverb 2 for all my simming, it's basically impossible to move back to "flat" screens again. It gives you a completely different depth-perception that is as vital when you fly as when you race, so basically any simming is a lot easier and more fun with VR. But again, if you make the plunge into VR simming, it's short of impossible to go back to "normal" afterwards.
> hard reality is it's not going to make its way into mainstream consumer everyday use this time either.
Yeah, simultaneously I agree with this. VR-in-motion (so not sitting still) is still pretty bad, and the setups you need for good performance are pretty expensive, so it's unlikely to break into mainstream unless some breakthrough is being made. With that said, there are niches that are very well served by VR and personally I guess I hope it'll be enough when the mainstream hype dies off.
IIRC there was a brief VR spike around early 2000s. I remember trying out Duke Nukem in a helmet and a three-button controller.
And then a bit later, there were 3D glasses, ones that synchronized with the high-rate monitor to show each eye its respective right and left frame. The demo for it at the time was Rogue Squadron and I thought the effect was amazing.
AR is the true future, but we're a materials science breakthrough away. You need waveguides or some similar thing that generates holograms that's cheap, has a wide FOV, and works in bright light. HoloLens and Magic Leap came close, but people couldn't figure out how to make enough money off the devices, apparently.
Maybe the change is the internet amplifies those holding the flame longer, like those responding parallel to me
I think the summary of the video captures this essence really nicely.
> If your idea has been tried before, do your research and understand why it didnât work. Assume the founders who tried before were very smart, very determined people; whatâs different now?
If you can't answer that question, don't try again.
The video has a good heuristic to apply that I think works even within changing contexts: "avoid things with a high supply of founders who want to work on it but zero consumer demand for the thing itself", the classic one being a discovery/recommendations app.
Discovery and recommendation apps for music and videos, such as Spotify, YouTube, and TikTok, are big hits.
You just have to have a colossal inventory, and a reasonably good algorithm.
Discovery and Recommendation in those apps seem to be features, not the purpose of the app itself.
You can't have a discovery app _that doesn't host the content_.
The main purpose of these apps is to click on them and watch/listen to some $foo. Not to just tell you what to watch/listen to.
In central Europe our biggest tarpit are sustainability / climate topics. Even founders that are smart enough to realize the difficulties still pursue these topics because of an unfoly feedback loop where government agencies almost exclusively fund those "societally important" areas. There's no other risk capital available so most founders align their goals, just to fold 1-2 years later.
I'm not sure I agree with this - though I'm not in central Europe.
Sustainability and climate are not products or businesses in themselves.
If you're trying to sell sustainabiity, what is your business? What are you selling? If it's advice on how people can be sustainable, or carbon credits or something, sure, those are things that may be tarpits.
But electric vehicles were probably a tarpit idea until they went from being advanced golf carts to being real cars.
Environmentally friendly packaging is doing very well. Etc etc.
I mean that's better then funding a bunch of startups that try and use psychological traps to capture the attention of your kids and ruining their ability to focus. Most of them also fold 1-2 years later.
Yeah this is actually a really useful idea in general.
I learned this when I was put on a team for a modeling competition in college. You had like 72 hours to solve the problem and write the report. It was really stressful but a LOT of fun.
There were a range of topics that you could choose from. Some were really obvious how to apply mathematics to, and some... weren't.
I was the math talent on the team... but my team members talked me out of going for one of the problems that were easy to apply math to. We instead picked the problem where it was LEAST obvious. And... we ended up winning the competition against a field of 10,000+.
I think that lesson applies in business all over the place. There's actually a lot of good comfortable money to be made in unglamorous industries.
I donât think thereâs zero demand recommendation apps, a lot of founders choose this because itâs a problem they want to solve for themselves, and there are a few success stories out there. Itâs just that itâs a super-hard problem
"It's a problem they want to solve for themselves", but note that they haven't tried all the alternative recommendation services and are only creating one as a last resort. They want to solve the problem, which is a different drive from wanting a recommendation app.
Now, if someone made a "Recommendation-Engine-in-a-Box", where someone who wanted to make a recommendation app for themselves would supply the content and could tweak the algorithm and the design, I could see that being successful in this market :)
> Itâs just that itâs a super-hard problem
I spent 2024 building an awesome TV series recommendation platform. It worked by matching you to professional critics who shared your tastes, by basically crawling Rotten Tomatoes and getting an LLM to grade the reviews out of ten. The recommendations were awesome, and having a personalized Rotten Tomatoes where you could read about and research the show using reviews by people who felt the same way as you did about stuff was freakin' cool.
However, getting people to actually sign up and use the app without a massive marketing budget was very, very difficult. The stickiness to get people to go back to it is difficult. Asking people to input their preferences in the first place is hard. People also simply didn't believe the recommendations, and wouldn't take chances on shows; the computer can recommend The Detectorists to as many people as it wants, but there's a high number of people who would love the show but will dismiss it looking at the cover image and having a quick read of the synopsis.
The recommendation part isn't super hard, the getting people to use a B2C app is super hard.
Whatâs a recommendation app? Like, Iâd like to watch this movie, can you tell me if it is streaming on anything?
Yeah, what was Zomato if not this?
Similarly a lot of people give up on something in their life, e.g. finding a partner, because of earlier attempts in different (worse) conditions.
Quite often we are incapable of identifying what those different conditions were. When something you don't think is important is the actual cause of the failure you're unlikely to notice and misatrubute the cause.
I used to be a schoolteacher, so whenever I read about someone's shiny new EdTech idea, I can't help but think that it's a tarpit idea.
Every developed country has a set of professional standards for teachers, and teachers who don't live up to those standards are pushed out, sometimes by having their teaching accreditation revoked. In Australia, for instance, there's a set number of hours of 'professional development' that teachers have to do every few years, and if you don't complete them, you lose accreditation and have to find a new career. The professional development activities and courses that meet the requirements are audited by the Department of Education, and have to draw on the latest research in educational psychology: keeping up with the latest research is the entire point of that professional standard.
When I did my teacher training, the first thing we were told in the first lecture was to never cite any research older than ten years, because it would be out of date. Now, if you've trained in the sciences - I was a physicist - you should be troubled with this, because a discipline can't really accumulate knowledge about the world if it throws everything out after ten years. That's why, when I broke the rules and searched through the databases of academic literature going back more than ten years, I saw the same ideas being reinvented under different names in different decades.
So there seems to be a bit of a trend for people to build flashcard-type tools at the moment, probably because someone's seen a paper on spaced repetition. That's nice, but you can't build a business around this. It doesn't matter if all the thought leaders are all in on spaced repetition this year, because next year they'll have moved on to something else, because they need to have something new to talk about. In Australia and the UK at least (I don't know the figures for other countries), half of all teachers leave the profession within five years of joining it, so most of your user base is overenthusiastic twenty-somethings with no life experience (yes, I was one of these) who will do whatever The Research tells them, and the ones who stay long enough to gain leadership positions tend not to grow out of this, so the classroom side of EdTech is basically a bunch of fads, so it's impossible to build a stable business in this space.
If you want to sell software to schools, go and work in a bunch of them, find some obscure administrative problem, and solve that.
Interesting. I don't mean to detract from your main point but as someone who's deeply into spaced repetition for the past 5-7 years (Daily Anki user + built my own spaced repetition systems for learning various skills), I find myself disagreeing with you on some things that you mentioned:
1. Spaced Repetition is not a fad. It's the most consistent and reliable way we we know for rote memorization (conditions apply). And it's not a new thing either. It's been around since the late 1800s. It just wasnt practical until the advent of computers and mobile devices. So I'm skeptical that there is another "something else" to move on to that is as impactful as SRS.
2. Not sure what the state of education in Australia these days but speaking strictly from my school days in India (1980s-1990s), something like spaced repetition would have been a godsend for every single student. And I'm 100% sure a vast majority of schools and teachers still havent heard of it.
3. I've been learning German for the past few years from some of the top private institutes in Vienna, Austria and let me tell you that neither the teacher, nor the students have any idea about spaced repetition.
That said, you're probably right about the business-viability of such ventures because of the difficulty of selling to the decision-makers, I just strongly disagree about Spaced Repetition being a "flavor of the year"
Sure, but there are all sorts of pendulum shifts in the teaching zeitgeist. Spaced repetition was just an app-based example: it's been a few years since I was in the classroom so I'm not sure what the flavour of the year is, but if spaced repetition were currently popular amongst schoolteachers, the swing away from it that I would expect to see would be to argue that memorisation isn't really learning and that learning experiences should be about developing deeper understanding, and so on. There is no measure of 'best practice'; a lot of these shifts are driven by personal preference.
> argue that memorisation isn't really learning and that learning experiences should be about developing deeper understanding, Agreed, but if you ask anyone who's SR for any amount of time will tell you: It's realllly hard to be effective with it if you don't understand the underlying concept. The order of operations is "first understand, then drill". Of course, this comes with nuance. There are things that just have to be drilled and others that don't even need any drilling if you understand the concept. And I'd expect those educators to know the difference.
> There is no measure of 'best practice'; a lot of these shifts are driven by personal preference.
Again, you're probably right but using the example of SR threw me off because it's the one thing where I think the data is so clear that it's easily justifiable.
I didnât read GP as making a judgment that spaced repetition was new or a fad, but rather that in the environment of how education decision-makers shift focus to new things, itâs a current flavor-of-the-year.
We saw elements of this with ânew mathâ, Singapore math, and common core math, each label of fairly similar concepts promising to improve kidsâ facility with math. Test scores havenât leapt though.
Yeah, you're probably right. I agree that the problem might be shifting focus every year but the one difference with your math examples is that Singapore Math/Common Core math etc. all seem like different systems that don't build on top of each other. You (I'm assuming here) can't focus on Singapore Math one year, then the next year to add Common Core math on top of that etc. Its one or the other.
Spaced repetition on the other hand is a cross-disciplinary technique that just needs to be introduced and kept there. There's nothing else out there to substitute it with. If the young staff hype it up one year and then it becomes part of the curriculum and then they move on to other fancy edtech things, then there's nothing wrong with it.
I partially agree with what you just posted, but â walking along your train of thought, I take a bit of issue with the following paragraph (sliced for emphasis):
> So there seems to be a bit of a trend for people to build flashcard-type tools at the moment, [...] so it's impossible to build a stable business in this space.
I am of the radical idea that lots of things should not be for-profit businesses (doesn't mean that it can't be profitable â just not exorbitantly so), and that economist's mistaken goal of exponential growth expectations is criminally separated from the sigmoid limits to reality.
So, therefore, while I agree that EdTech is a bunch of fads, I think the fact that EdTech is a thing is wrong.
And I agree with your main point that we should be chasing accumulation and refinement of knowledge, and not doing some sort of spring-cleaning every 10 years.
"EdTech" is ripe for disruption - by a non-profit, open-source entity that provides "school stuff as a service" but is basically a lifestyle business.
It would have to be funded by adventure capitalists (e.g., retired techies having fun building stuff) for awhile, but it could easily take over once it got traction.
> If you want to sell software to schools, go and work in a bunch of them, find some obscure administrative problem, and solve that.
I think this principle generalizes to: If you want to sell software to X, go and work in a bunch of X, find some obscure administrative problem, and solve that.
Although some people seemingly have a talent for selling products into industries they have no idea about. I always assume that means a highly motivated buyer.
There's also a different angle: EdTech doesn't have to sell to schools, but could also be learner-facing.
The tarpit idea is very descriptive in hindsight. What makes something a tarpit is an idea that sounds cool on the surface and is accessible (donât have to be one of a kind founder to do it) and when you talk to your friends and people who might be customers you get very positive feedback. So it all starts to feel like a slam dunk. However, if you are a VC you will have seen this exact idea or close variations in it a hundred times and they all flamed out to a zero. As the VC you have visibility into common failure modes ( not able to charge enough, no scaled market, not sticky enough, etc) what is hard from the founder side of things is all those issues and many more are common to almost any venture until you crack the problem and get market fit. So the tarpit concept is more a description of VC scare tissue than a fully operational definition for founders ,because a former tarpit can become a blue ocean of opportunity ( uncontested market) if some element of the equation changes ( technology shift, culture shift, deep founder insight etc)
So as a founder how can you tell if you are about to jump into a tarpit?
1) do a lot of research on the problem and see what has been done in the space in the past and who is working on the problem now. If you find lots of failure - dig in and try to understand what the core failure modes were. 2) work on something that people will pay you for, even a very ugly early product. Income is a strong validation. 3) reconsider your idea if it requires the incineration of mountains of cash to get peopleâs attention.
But at the end of the day Tarpit is really a descriptive heuristic that VCs can find to be useful but not absolute.
I found imagining the actual metaphor of tarpit enlightening - it looks like a nice healthy pool of water but turns out to be sticky and impossible to get out of. Under that attractive surface are all the corpses of everyone else who charged in and got stuck.
So itâs exactly like you say: you keep your distance and you look for evidence.
I do think the point of the metaphor is that sometimes a tarpit is just what it is. I.e. there is no value under that shiny surface and youâre only getting stuck staring at it.
An idea is tarpit until someone, or some new tech, or regulation cracks it.
YC has a rare opportunity and it squanders it. It is a hub that gathers most problems and approaches to them in each discipline and many many failures on them at least once a quarter and all of that goes down the drain, instead of being published and explored publicly. The energy of bright new founders is not spent re-hashing the old but exploring the new. YC can still evolve into a science hub for things people want with much more impact than it does now. New founders want to protect IP and hold back competition, so publish the failed ideas and approaches - make it a competition. Show the full length and breadth of tarpit zones and any time they may be cracking. This way new energy goes towards better VC returns instead of falling into old cracks. Build a Yelp of things people want that need to be built or solved.
I think an underrated aspect of this is also that YC is ultimately a VC fund and so they're talking about companies that have the potential to be massive, multi-billion dollar companies.
Many typical tarpit ideas (to do apps, habit trackers, note taking etc.) can be great businesses for a couple of people building software together but not have venture-scale outcomes.
I do agree that as soon as you get network effects (recommendations, marketplaces etc.), SOOOO much is tarpit.
B2C is virtually impossible compared to B2B. This may not be immediately apparent but it is so obvious in hindsight.
The biggest reason I think founders are going for B2C is because they have zero clue about how to network and sell to other businesses. It's easy to set up a shopify account. It's hard to cold call your first prospect. Do you even have any prospects? Do you know how to find them?
The advantage of B2B is that once you figure it out for the first customer, you are on an exponential path to happiness. You can practically cancel your marketing budget at that point. B2C requires an ongoing assault on the dopamine economy. Unless you can get someone on a subscription and program them to forget about it, you're gonna get steamrolled by TikTok & friends.
B2B puts you at the mercy of "your next customer's wished feature".
The bigger the "B" you're trying to sell to, the bigger the desire to say yes. This is a very dangerous path.
If you're DHH, you can say no. If you're responding to RFPs for a living... It becomes complicated. Not impossible of course, but it's a different path altogether.
> B2B puts you at the mercy of "your next customer's wished feature".
Yes.
> The bigger the "B" you're trying to sell to, the bigger the desire to say yes.
Absolutely.
> This is a very dangerous path.
You lost me here. What you term "a very dangerous path" I call "a roadmap".
The bigger the customer, the more they can pay. Also, just because it's 'B' and not 'C' doesn't mean there aren't humans involved. You can go a long way (often, all the way) with a simple 1:1 conversation with the most senior person responsible for your project on the customer's side.
> because they have zero clue about how to network and sell to other businesses. It's easy to set up a shopify account. It's hard to cold call your first prospect. Do you even have any prospects? Do you know how to find them?
How do you learn this skill? Any resources or books you recommend?
Sure, a lot of people have written about this but this isn't a skill that you can learn from a book.
Which is why it feels virtually impossible to many founders. Some people may claim B2B is easier, but only if you have the skill and it is hard to figure out how to obtain it
My experience is the complete opposite. B2B is almost impossible, even when you have a great product for a great price. B2C on the other hand is a delight, with much better prospects for growing your clientele.
The only B2B ideas I have would apply to one company i'm familiar with, maybe a couple of similar companies in the same space. How is that better than B2C?
How big is that space in terms of businesses? Can you reach them more easily than you could reach a similar $ value of customers? Can you expand your reach and learn more about the market once you're embedded? Are they easier to sell to than consumers? Can you sell them support or maintenance contracts?
etc...
I guess the logic behind B2B vs B2C comes down this belief: It is easier to sell one unit for $1 million than one million units of something for $1.
The video has a sequel released nine months ago: https://www.youtube.com/watch?v=GU9iT7MW0rs
And for completeness, here is a video from 2022: https://youtube.com/watch?v=Th8JoIan4dg&t=195s
Some ideas are tarpit ideas until enough people get stuck.
Location estimation (figuring out where you are) based on indoor WiFi / BLE is one example. Compared to 15 years ago, we have (IIRC - I don't work in this space) super-precise timing API from the modem, and there has been work on the reflections issue (the two big problematic things that non-RF people typically miss).
I did a project on this in like 2010 as a student hobby project. It wasn't accurate, but I also had no idea what I was doing. I mostly did it in a naive way, where I mapped out signal strengths in various rooms (it was dorm floor, 2 rooms shared each) and then trying to figure out where I was based on it. As a non-CS student I thought it was cool......
It was cool
I think there is a difference between ideas that are based on a fundamental misunderstanding of how the world works and ideas that we do not currently have the capabilities to solve.
What about multipath? Is that not an issue at wifi wavelengths? Or is that a sub or superset of reflections? It was quite a long time ago now that the first "proof" that one could use leaky WiFi to "see through walls" and observe people moving around inside a building from without.
Participate in a transmitter hunt (also called a foxhunt or a t-hunt) where the organizers or the people hiding the transmitter know their stuff. Reflections and multipath can lead you miles away from a transmitter location.
Anyhow, someone asked me if I knew how to do this without consent once; that is, if I knew how to track people in a building without them knowing. This was 8 years ago or so. I had hoped saying "that's not possible" would dissuade them, but instead they just never spoke to me again.
Oh yeah, multipath is the term I was thinking of. As I recall, the issues were that a "longer" path may have a stronger signal than a more "direct" path, and that a straight line of sight path is not a given.
Get a daily email with the the top stories from Hacker News. No spam, unsubscribe at any time.