So much of VC funding is just about being in the right place at the right time. A lot of these guys seem like geniuses because of winning bets, but that's because nobody talks about their flops.
It's entertaining watching all of the VCs who have made a major bet on crypto, who are now doubling down on what looks more like an overheated casino losing steam. Fred Wilson might end up watching the tide go out on this one. Maybe people will finally remember that he was part of Twitter's clueless board, the one that didn't think it was important to buy Instagram?
>, but that's because nobody talks about their flops.
The KPCB flops in "green tech, clean energy, etc" were widely reported.[1] Also, the VC firm DFJ and Tim Draper in particular were laughingstocks for investing in the Theranos scam because they didn't do proper due diligence.
>Fred Wilson [...] was part of Twitter's clueless board, the one that didn't think it was important to buy Instagram?
Is there a source that says it was Fred Wilson who blocked the acquisition of Instagram?
The story I read was that both Jack Dorsey of Twitter and Mark Zuckerberg of Facebook were courting Instagram at the same time.
Jack offered ~$500 million (all in Twitter stock).
Mark offered ~$736 million ($300 million of that in cash).
Basically, Facebook had the more enticing offer.
In 2012, Twitter wasn't generating any profits and didn't have any excess money in the bank to include any cash component to their bid.
On the other hand, Facebook was already profitable for 3 years and had ~$4 billion in cash in the bank even before the IPO in May 2012. Having a stock that was backed by real profits and a war chest of cash lets Facebook make more attractive acquisition offers.
Do you have any sources for the DFJ/Theranos backlash? I've wondered if that would hit them, so I'd love to read more if you're aware of any good journalism on that specific investment by DFJ.
This comment could benefit from a little more research. Sure, being a successful venture investor is about being in the right place at the right time, but USV is one of the best out there. Fred Wilson has written extensively about his flops and misses (they passed repeatedly on Airbnb for example), and has also written extensively about the dynamics of investing across risky-enough ideas that there are a lot of failures in a successful portfolio. Sure, Twitter has made a lot of ridiculous decisions, but as an early-stage investment, it returned the fund.
My take is that, while there is skill involved in picking venture investments, success in the field has more to do with social skills, status, and wealth than it does with skill. Only the richest have access to the deal flow of the best entrepreneurs.
The success metric in VC is financial return (IRR or cash proceeds / cash invested). % of investments that succeed or fail is not really that important. The VC model is based on finding a few really big investments at the cost of many that don't work out. Something like 60% of VC returns come from less than 10% of deals
Sure, but when vetting a baseball player you check the batting average; when handicapping a horse you check the races they've lost for conditions they can't handle as well; when comparing Halo player stats you check the kill/death ratio.
Vetting a VC only by wins is missing quite a bit of information, presumably. It can at the very least tell you about potential blind spots and some idea how much actual skill may be involved versus luck or even "a broken clock is right twice a day" scattershot portfolios.
I think the analogy breaks down somewhat due to the relative scale of their misses vs. hits. A batter can only get so many runs with one hit, but a good investment can make all of the misses look like rounding errors.
A batter can help win a series, especially the World Series, and make all their losses look uninteresting. But you still are going to at least check their batting average if you are looking to trade them.
A horse can win a major stakes race, such as the Kentucky Derby, and bring in a massive purse (and eventual stud rights) that makes any losses look like warmup rounds. But before heading to the betting window you are still going to check if they've lost any big rainy races if it's raining on Derby Day.
A Halo player can win a championship or tournament and no one will question any losses they've had. But you still are going to want some idea of their kill/death efficiency before scouting them for your team in the next deathmatch.
And batting .300 is considered good -- high rate of failure doesn't necessarily mean something is bad. The metric VCs are judged on is financial return. If they invest in one company that gets them 200x return and 20 that lost all money, they've still done well at their job
I'm only arguing that the stats matter if you are trying to compare VCs, establish baselines of VC ability. I'm sure that they might be skewed, but that doesn't mean they aren't potentially valuable information.
That .300 is considered good today based on the current environment of the sport and in comparison among peer groups. That same .300 looks shabby in a previous era or in particular sub-leagues or among particular types of batters (your benchmark for designated hitters might be higher given their only focus is hits, for instance).
Maybe it wasn't important at the time. Instagram only became wildly popular after they introduced stories, which they didn't have at the time, did they?
I think OP's larger point stands -- it might've played our differently had Twitter acquired instagram. Maybe they would've done what they did with Vine.
So after browsing through Blockstack's website and a few of these Dapps there's something I don't quite understand. Storing data into the blockchain itself is obviously extremely expensive so I wondered how they did that. Turns out that for most of these apps they either don't tell (as far as I can see) or have this exact same snippet of text in their FAQs:
>By default, your data is stored in a dedicated Microsoft Azure Blob. But you can and should connect your Blockstack Browser to your own cloud storage solutions (preferably multiple).
So... what's the idea here? I don't need ethereum to encrypt my data and upload them to various cloud storage services. Sounds to me like blockstack is an app framework like Google Play or the App Store, only you pay with ethereum instead of fiat and your data is encrypted locally before being uploaded. I mean that last bit is definitely a good thing but where's the famous "blockchain technology(c)(tm)" exactly?
Ah, thank you for confirming that, I didn't think about looking at Blockstack's FAQ itself. But then I reiterate my question: what does the blockchain enable here exactly? TFA says that it's "exciting for crypto[currencies]" but it just seems like a cloud computing framework that happens to use ethereum for purchases.
Just to be clear I'm not saying it's useless, actually being able to host your own data anywhere you want and having "end-to-end" encryption sounds great, I just don't see the "blockchain is going to revolutionize the world" angle.
Part of the problem here is that we are still in the infrastructure phase of blockchain development, so most dapps still rely on multiple centralized services. But as the infrastructure develops that will shift. For example, (i) all of the blockchains are working on scaling, (ii) Sia, Filecoin and a few others are working on blockchain based storage systems, (iii) Maker Dai, Basecoin and others are working on a stable coin, (iv) 0x and others are working on decentralized exchanges without custodians, (v) many are working on making addressed and key management easier, (vi) distributed identity verification and 'user trust' systems are developing, etc. All of that will take time, but as systems develop, more complex and more truly distributed dapps will be enabled.
Joy stream appears to be a service which rewards people for hosting torrents. I could see this being used by a lot of people.
Are their any services that provide usage figures for any of these Dapps? I’d be curious to know how much they are used as I think Dapp usage is a good bellwether for utility token valuation.
I really want to see crypto currency move beyond just being used for speculation.
The Dai stable coin. Many different smart contracts are constantly working together to keep the token worth very close to 1 USD: https://coinmarketcap.com/currencies/dai/
The core mechanisms are enforced by smart contract, dai issuance comes from multiple entities, it provides (or aims to provide) a large number of users some utility (stable purchasing power, whether to hold or spend) - why is it not a dapp?
In general, it is better if the value of your money that you want to use to buy things does not fluctuate every minute like in the case of Bitcoin.
People are working on Dapps for insurance, lending, prediction markets. It is not practical to take out a loan that is denominated in a volatile currency. Also, If you want to bet on the next president in 2020 for example, it is not practical to make the bet using a volatile currency such as Bitcoin.
USD is not the ideal currency in terms of value stability. I know nothing about Dai, but if it works, its technology could be expanded to other, more material, more reliable assets than USD, and that would be great.
USD can't be sent over the wire without the approval of and fees paid to banking institutions, with which both parties must be previously related and approved by.
I assume because it allows you to trade in and out from crypto positions to something that tracks fiat.
This of course assumes that this coin can actually hold its value in the event of a market crash. I also assume it’s not backed by real dollars held in a bank but some kind of basket of other cryptoassets right?
It's currently backed by ETH but is moving towards multi-collateral assets. Today, they announced a partnership with Omisego [1], with one goal to allow OMG to be used as a type of collateral. There's also chatter about using DIGIX, which is another stablecoin pegged to the price of gold.
The mechanism by which they maintain stability is quite fascinating. From the whitepaper [2]:
> The Dai Target Price is used to determine the collateral-to-debt ratio of a CDP, and thus the Target Price represents the price at which Dai is backed by collateral in the long term. The Target Price is continuously adjusted according to the current Target Rate. Automatic Target Rate adjustments ensure that the Dai market price remains stabilized around the Target Price in the short term.
> When the market price of Dai is below the Target Price, the Target Rate increases. This causes the Target Price to increase at a higher rate, causing generation of Dai to become more expensive. This leads to CDP users covering their CDPs and leaving the ecosystem, causing the outstanding supply of Dai to decrease. At the same time, the increased Target Rate causes the capital gains from holding Dai to increase, leading to a corresponding increase in Dai demand. This combination of reduced supply and increased demand causes the Dai market price to increase, pushing it up towards the Target Price.
No it is not backed by real dollars in a central location. It's a kind of a derivative. The way it works is very interesting. Recently Andreessen Horowitz and others invested 12M into the project.
Anyone else a programmer/tech person that thinks blockchain is awful for storing information?
Centralized servers are extremely good at what they do. I feel like most of the time blockchain isnt needed unless things need to be decentrally verified.
I was considering doing my own 'cryptocurrency', but I decided a centralized server was better for doing the job at hand.
It seems like a really good way to store small pieces of information that aren't changed very often. Such as Name System Registry entries. It works really well for this type of use case, and name registries are good things to be distributed. Using an EVM based blockchain allows me as an engineer to get strong identity, authorization and storage guarantees for free, leaving me to focus on the logic and systems around how names are changed and what those names mean.
You might have had misconceptions about the blockchain, but you seem to have come to the correct conclusion. It is awful for storing information, and it's generally not meant for storing information. Obviously you can combine traditional servers and blockchain in a such way that the system is completely decentralized, such as Sia https://sia.tech/.
Patchwork looks great. But it's an old-school distributed/federated app in the vein of IRC or XMPP, right?
The definition of a "dApp" these days seems to require blockchain, either for storage or for a token ecosystem. This co-opting of the word "distributed" is unfortunate but probably as difficult to fix at this point as the misuse of "crypto".
Dapps mean apps on blockchain and consensus. Patchwork does not apply, otherwise one could mention "Torrents!" etc. You would be at least 20 years late to the party.
> Are their any services that provide usage figures for any of these Dapps?
The closest I've seen to this, eg. an Alexa.com/AppAnnie for Dapps, is https://dappradar.com.
It measures DAUs, transaction counts and daily/weekly tx value (in ETH) across many Ethereum smart contracts. It's not perfect but does provide a decent standardized view of some Dapp usage/traction.
In the future it will be possible to use Ethereum Dapps even if you don't have any ETH. The way it works now is that ETH is needed in order to pay for gas. However, it will be possible for the contract to pay for gas in the future. Developers would be able to add some ETH to their contract, so that it is free to use (or some parts of it are free) if they wanted to.
I'm also on the skeptic side of things though this intrigues me because it perhaps provides an alternate model to online ads which we're finally discovering the negative side effects of.
Isn't it integral to the way these dapps work however? Is there an other way to achieve the same results? After all eventually somebody has to pay the bill for there computations.
It seems to me that the big problem of dapps is not so much the fact that you need to pay to deploy them but that you have to pay so much to do anything meaningful. The system is extremely expensive compared to regular "not-trustless" distributed/cloud computing offerings out there.
For instance while browsing through the blockstack website I saw that the first such application was a "docs suite" (à la office/google docs) called "graphite": https://www.graphitedocs.com
That seemed interesting but my first thought was "man, if they store the files in the blockchain that must be absurdly expensive to use". They probably thought the same so, according to the FAQ at https://www.graphitedocs.com/faq :
>By default, your data is stored in a dedicated Microsoft Azure Blob. But you can and should connect your Blockstack Browser to your own cloud storage solutions (preferably multiple).
Oh. So that makes sense but doesn't that destroy the whole point of it? What if I decide to, say, use OpenOffice on my computer, encrypt the files using GnuPG and then upload them to dropbox, spideroak and some AWS bucket, wouldn't that effectively grant me the same privacy and control over my data?
I'm going to sound like a naysayer but it just looks like yet an other example of using blockchain for the sake of saying that you use the blockchain.
But of course it's just one dapp out of many. Let's see the next one, "stealthy": https://www.stealthy.im/
It's a "A secure decentralized communication platform". Now storing small text messages in the blockchain would still be very expensive but it might be acceptable, especially if you worry a lot about censorship or your message remaining available "forever".
>By default, your data is stored in a dedicated Microsoft Azure Blob. But you can and should connect your Blockstack Browser to your own cloud storage solutions (preferably multiple).
It adds an extra hurdle to developing dapps as well.
So many users in gitter channels I frequent run into problems with gas.
But one thing is amazing is that once your dapp is deployed your backend is fully permanent and replicated 19,000 times [0] (which is both super good and super bad at the same time - no bug fixes in prod).
> The bottom line for me is that we are finally seeing some useful decentralized applications being built for consumers on these blockchains.
I don't know if the word "decentralized" should be used so merrily. Research was published recently [0] showing that the blockchain protocol requires a central authority to prevent coordination failures and externalities. Maybe these "useful decentralized applications" should come with a caution sign to the consumers: "product will not only be slower and more expensive than the centralized alternative, but there is also risk of coordination failure and double-spends".
The same tension applies to a hypothetical mesh-network Internet, compared to the telco-backbone Internet: good for freedom and autonomy, bad for usability and performance. Alas, consumers tend to vote with their dollar for the latter. The "siren server" problem described by Jaron Lanier in "Who Owns The Future?" is a wicked problem when it comes to marketplace traction.
In the case of both mesh networks and dapps, I think the greatest potential is in the developing world, where centralized infrastructure is absent or inadequate.
I saw the exact same line and I wanted to take out the word "useful". Don't get me wrong, I am as excited about Dapps as the next person, but I have not seen use cases that solve existing problems.
What is the "blockchain protocol"? I think we really need to pick A blockchain and criticize it specifically.
And with applications: some will be more expensive, some won't be. Ethereum developers are already exploring ways for dapps to pay for gas instead of users (searching for link).
They define what blockchain is in the first two sections of the paper. Their model of the blockchain applies to both Bitcoin and Ethereum, for example, and seems general enough to apply to most networks people today refer to as "blockchain".
I investigated “Dapps” the other day, and I’m struggling to see the point of these. Yes, they use a blockchain to store their data, and their data manipulation logic (“contract”, or schema, and stored procedures as we’d call it in the RDBMS / Monolith days). The UI, and such is all run by some centralized entity. If the App needs cryptocurrency in order to work, you have to run your own wallet.
If (1) I need to run my own “geth” daemon, and (2) some central authority’s ownership is reigns over the apps availability, I don’t see how this is better than the current world. Until there is a Coinbase-like experience to solve (1), I think it’ll be very difficult to get real market adoption. (2) to me strikes a whole set of problems — What’s the point of Dapps? I would think it’d be better if the UI and API was hosted on a blockchain, and there was some kind of UI (browser?) with an embedded wallet, and agent to be able to interact with contracts. I know some of this is in development, but is anyone close to the ability to run a Reddit-like site without the existence of some central authority, along with being economical to run?
You can host the UI wherever you want. You can put it on IPFS, you can broadcast it as a bittorrent, you can put it on Github, you can put it in a pastebin, you can even put it on the blockchain.
A lot of developers choose to have a replica hosted on traditional HTTP for easy onboarding, but the point is that the application itself can be completely decoupled from the developer.
Even if the developer hosts their app on their own domain, there is nothing stopping you from forking it and replicating/hosting it elsewhere. It's like the ultimate extreme of open source, but also applied to APIs that can't be taken away.
For example, you can launch a smart contract which hard-codes an IPFS hash to a GUI that can never be changed, even if the original owner wanted to.
It's more about what can be done, rather than what is being done today. The possibilities are inspiring and attractive.
> But, if you put it on IPFS, what hosts the wallet, and wallet API?
Which wallet? You can use whatever wallet you want which supports the functionality the DApp requires. There are dozens of wallets, some are in-browser.
While I didn't mention anything about censorship, I suggest reading that entire thread and its continuation. All it discusses is the ability to have curated blocklists. It's up to you whether you want to use them or not.
What I did mention is that you don't have to use any specific replication protocol. Don't like IPFS? Use something else.
If you did want to use IPFS and you're concerned about censorship, then consider that IPFS is a distributed hash table protocol with some default client rules. If you don't use their default bootstrapping nodes, you can create your own IPFS network. In fact, IPFS is totally usable in an isolated local network, or even over a sneaker net.
We have embedded wallets in the browser with things like MetaMask, which lets you execute Ethereum transactions by clicking buttons, fetching data from the blockchain to display on the webpage, etc.
Dapps are still evolving. Some offer utility from fulfilling niches that are vulnerable to centralized power; ie. a decentralized Youtube that won't take down your video. Some are safer/cheaper implementations: ie. Hexel (YC) a platform that lets people create/manage tokens, like if you wanted your community to be able to send/use tokens.
A mistake is assuming you need to put everything on the blockchain, and a new product should be fully decentralized. But in most cases, you can still deliver utility by having only the important pieces on the blockchain.
Metamask is quickly becoming one of the most interesting projects to me. It's so easy to spin up a site now and use metamask for identity and authorization. If you have API's that need to be protected from denial of service attacks, you just channel them through a smart contract and make the user's pay the gas, no more need to protect your backends.
The blockchain just becomes a source of truth and a protection against DDoS attacks. Past that, all other "transactions" can be done off chain. Future is exciting!
I understand the frustration, as someone working on apps in the space it is clear that there is much to be done.
Dapps right now are still a work in progress and very early. Most of the issues you brought up are known in the space and are actively being worked on. For example, UI file storage is being tackled via the work of ETH Swarm + ENS (also Mist browser), EOS Storage, IPFS Filecoin, etc. Running a full node is not necessarily required, you could immediately access a decentralized network via a light client on many platforms.
Ultimately the end goal for Dapps are to merge the benefits of decentralization (always-available, secure, permissionless, distributed, censorship-resistant, etc) into traditional application models, and in some instances create entirely new types of applications.
As I've said before, when consumers speak of "the cloud" they are usually describing something that sounds much more like decentralized platforms than current centralized solutions. Long-term I think decentralized platforms are going to change every industry in some way. Even from just an accounting perspective it's hard to see how this tech won't change the way we manage and transfer funds.
> The UI, and such is all run by some centralized entity.
The UI has to be white listed, but dapps can be cached locally. There's a lot of argument about what is centralized and decentralized. It's quite a polarizing debate.
> If the App needs cryptocurrency in order to work, you have to run your own wallet.
To read from the app it generally won't require cryptocurrency.
> (1) I need to run my own “geth” daemon
There are public nodes (i.e. Infura) for Ethereum dapps.
> I would think it’d be better if the UI and API was hosted on a blockchain, and there was some kind of UI (browser?) with an embedded wallet, and agent to be able to interact with contracts. I know some of this is in development, but is anyone close to the ability to run a Reddit-like site without the existence of some central authority, along with being economical to run?
Toshi is developed by Coinbase. There are a few other projects as well like Status. It's very early days though. Blockchain typically has low TPS and the alternatives are very complex. I'm hoping technology like Hashgraph or Hashgraph-like will deprecate state channels and payment channels.
I would think that, depending on the dapp, it would be inaccurate to call the UI centralized, if you were able to run it yourself. In a way a wallet is a UI for a currency dapp, and you could either use a centralized UI or your own instance.
They controlled 26.6% of the total Ethereum hashing power.
So, lets say this twitter clone - Peepeth takes off. And then there is someone who wants to post some stuff against F2pool. What is stopping F2pool from manipulating the feed? You might say other miners controlling 75% of the hashrate will pick it up. Sure, you might be correct. But what if the post shits on every other mining pool out there? What happens then? What if the post is time sensitive?
Decentralized apps can only go so far considering there has to be a record for the system to work. And if record keepers manipulate the system it will fail. The question becomes then:
> And then there is someone who wants to post some stuff against F2pool. What is stopping F2pool from manipulating the feed?
It's the difference between 'unstable equilibrium' vs 'stable equilibrium'.
The power of a centralized entity on their service, like twitter on a feed is a stable equilibrium, they can modify something and not face too many repercussions (like when we see what happened with Wells Fargo scandal).
On the other hand, the power of a major player in a decentralized setup is an 'unstable equilibrium', if you modify something into your favor, your power goes away really fast. And cryptocurrency networks have shown that they are extremely sensitive towards any changes or abuse of power made by a major player. It also makes a lot of sense from game theoretic perspective.
The hard fork made by the Ethereum community was such an example, it was so contentious that even though they succeeded, they are very resistant towards any such changes in future.
With Proof of Stake, the 'stability' of unstable equilibrium of F2Pool (presuming F2Pool is a large staking pool at that point) becomes even more unstable. In PoW if F2Pool performs a harmful action, then any attempts to undo that harmful action does not harm F2Pool, they just neutralize their attempt.
Whereas in PoS consensus, any attempts by F2Pool to perform a harmful action would result in them losing their wealth.
In most centralized entities, interests are conflicted. With mining, the value of the income is directly determined by the price of the cryptocurrency.
There are other incentives beyond most obvious ones. E.g. hijacked miner could attack to pump another currency (crypto/fiat) the miner holds.
Incentives are out of equation when you speak of fault tolerance. Incentives are not calculatable. Only "tolerance" (# of tolerable byzantine nodes) matters. And that's 2 for eth and btc. TWO.
Not exactly a Dapp, but to me the first killer cryptocurrency use case is/will be Althea: https://altheamesh.com/, incentivized mesh-networking and internet provider services with micropayments.
Hmm, I think one thing people forget, is that most of the time benevolent centralization, and revolution to another centralization when necessary is more valuable to consumers, than always on decentralization.
One reason is to move from being the product to being a user, so that your attention is not sold to the highest bidder and your behaviour manipulated in such a way that it produces the highest value. Of course, paying with your attention is a valid way of paying for things as long as the implications are clear.
IMO, federation is a better model of this. With Mastodon, sure somebody is still paying server costs, but there is no unavoidable fee for the software/service and it has the same censorship-resistent properties.
(I'd like to see signed messages in Mastodon but that's not a problem inherent to a federated model)
Censorship resistance. Twitter could delete your account (or some of your tweets) if they wanted to. They could also prevent your tweets appearing in other people's feeds. And they have done this to some controversial figures.
Basically a Dapp is a more neutral platform that costs money. Probably not worth the cost for the average person though.
If the price is meaningless amount I would actually be more comfortable paying than being a "product" that is being sold.
Especially if there is no friction in paying and apps would subtract little amounts of cryptocurrency and I could review from time to time if all is in check
It's important to get the relationship between IPFS and Filecoin correct. Saying "IPFS has a token called Filecoin" is not true, as IPFS works (today, right now) without Filecoin, which has not been built yet.
Filecoin is simply a way to store things in exchange for FIL (or vice-versa). Filecoin does depend on IPFS, but IPFS is a standalone project and does not require Filecoin to be useful.
Disclaimer: I work for Protocol Labs, specifically on the IPFS project
One project I really like is Streamr. It looks promising. Monetizing realtime data on top of ethereum. It's great to see projects on top of ethereum creating real value.
I respect your point of view. I like the project as i like Golem. They are trying to build decentralized cloud services on top of an existing blockchain.
Hacker news is not read by idiots, so please don't accuse me of trying to deceive readers. The average reader here is quite smart, they don't need your defense.
Fred may as well be describing the turboencabulator[0]. I get that VCs should swing for the fences, but this seems like they are totally ignoring first principles when it comes to cryptocurrency/blockchain apps.
What makes dapps "special" (i.e. why we have a separate name for them) is that they utilize a blockchain in one way or another. For example for storage or transfer of data.
I think the point is that's not actually an interesting feature. The programs use an over-hyped database hoping to siphon off a bit of the hype...did we have a separate name for programs using NoSQL too?
>have a separate name for programs using NoSQL too?
Yeah, horrible legacy projects that are a pain in the ass to maintain
Source: I am working on a legacy API based on a noSQL database with a very dumb design (because the data was relational of course) in a very large company
> Yeah, horrible legacy projects that are a pain in the ass to maintain
How can a "legacy" project be using noSQL? In my company, "legacy" projects are projects that have been around for 10+ years. MongoDB hasn't even been out a full 10 years, yet...
A project doesn't need to be 10+ years old to be legacy.
This API is 3y old, and was the first iteration, the v1, of the company's digital asset management API, it's only maintained for a few teams around the world, whereas the v2 is well crafted and has a way more logical datamodel
It's just weird when I read stuff like "Yeah, I've had to do a bunch of maintenance on our legacy webpack/Angular 2 system" because in my mind webpack and Angular 2 are still very recent technologies, just not recent compared to webpack 2/Angular 5
It's entertaining watching all of the VCs who have made a major bet on crypto, who are now doubling down on what looks more like an overheated casino losing steam. Fred Wilson might end up watching the tide go out on this one. Maybe people will finally remember that he was part of Twitter's clueless board, the one that didn't think it was important to buy Instagram?