It feels like Solana is moving fast and breaking things a little too much for a project that has to do with money. Uptime should be a competitive advantage for decentralized networks, but the Solana network went down 12 times in the first 6 months of this year[1]. Even Rust couldn't save them.

[1]: https://www.datacenterdynamics.com/en/news/solana-cryptocurr...

My understanding is that the outages are not bugs but a deliberate tradeoff between throughput and liveness.

No, they’re bugs, but Solana (average 4500 transactions per second) competes more with Visa (average 7500 transactions per second) and older blockchains aren’t even in the arena. The bugs get fixed, Solana’s pretty stable, you can see the exact numbers at https://status.solana.com/

Solana's definition of transaction is weird though. There's no way a P2P network is doing 4500 txs in the common blockchain terminology. The entire system is a house of cards waiting to crumble - just napkin the average tx size * 4500 txs * peers etc...

How are there enough transactions on Solana to get to 4500 per second? How is there possibly enough demand on a crypto network to be remotely in the same league as visa? I don't buy it.

Wash trading bots can create as much volume as you want.

A lot of Solanas throughput is consensus messages between validators.

But finance in general is many orders of magnitude more TPS than visa. Think about high frequency trading of derivatives of all sorts of things, and all the tiny arbitrages that can be done over all the exchanges when fees are cents. Those can easily multiply to thousands of TPS.

That’s literally what Sol was designed to do. Plus taking 20% of all NFT traffic in a year is a huge part of it. Even with consensus messages removed Sol still does more transactions than all the L2s combined.

But why is this the case when their volume is nowhere near that percentage of the L2 chains too? What exactly does the tx count include?

Got a reference for volume? Sol had either the most or second most active wallets.

Bitcoin can do that many transactions fairly easily with the changes we made for eCash. All the quadratic performance bottlenecks were removed.

I assure you that their status page is no indication of reliability. Ask the more serious node operators and they can tell you how frequently you actually encounter issues with the network.

Did it go down 12 times for technical reasons or was it intentionally pulled down? Something to wonder about.

Is there a difference? A decentralized protocol that can be DOS'd by a reasonably successful criminal gang isn't all that useful.

Doesn't sound very decentralized at all...

No it just had bugs that got fixed. Decentralisation does not imply perfection.

It goes down for an incredibly stupid decision. Transactions on Solana are free so spammers can produce millions of transactions per second and at some point validators simply crash.

Solana has subsecond transaction processing so a validator being down for more than a second counts as a downtime whereas with Bitcoin a block taking two hours to mine is completely normal. Bitcoin doesn't have the concept of "downtime" only the concept of delaying the next block.

> Even Rust couldn't save them.

Rust doesn't stop bugs, it just helps you try reason about stuff and maybe catch them earlier.

To be fair to the Solana team, their production chain is called Mainnet Beta.

Going forward their outages should occur less frequently since they've finally introduced a fee market akin to Ethereum's dynamic gas prices. With a dynamic fee market it will be more financially prohibitive to DDOS the chain.

I tried learning a bit more of the project and couldn’t understand the design decision and how it led to the claimed throughput numbers. Then I found Shoup’s take[1] and it all made sense. Highly recommend reading it.

[1]: https://www.shoup.net/papers/poh.pdf

From the paper:

> Indeed, a proof-of-stake protocol enjoys many of the benefits, in terms of efficiency, of permissioned protocols; moreover, while it is less centralized than a traditional permissioned protocol, it is not nearly as decentralized as a truly permissionless protocol, such as Bitcoin.

I hear this take often, but I have no idea why it would lead to less decentralization in practice. PoS means that someone needs to transfer you some of the native token before you can participate in consensus, but there's no practical way for a cabal of validators to stop someone from doing that if there's even one person with Eth willing to sell to you. That doesn't really seem different in practice from having to find someone willing to sell you a mining rig.

Solana isn't more centralized because of PoS, it's because it has extremely high validator requirements (including a 300Mbps internet connection, which rules out running one in most homes).

You could argue that PoW has natural attrition of validators because ASICS improve over time which gives more opportunity to enter the market, but that is secondary to the high validator requirements.

I used to run a Solana validator. After compiling the Solana perf libs for my installed version of CUDA Solana consistently used at least 75% of my RTX 3090. For a validator.

Solana validator hardware requirements are incredibly high compared to anything else I'm familiar with.

Have you seen what a Bitcoin ASIC miner costs?

Of course. Why (and what) are you trying to compare between Bitcoin mining (PoW) and Solana validating (PoS)? Five or so years ago I did about 25 kW constant load of PoW mining so yes, I'm pretty familiar.

In additional to the Solana validator from my original comment I've also run Ethereum and Cardano validators (or rough equivalents). They use a fraction of the compute required by a Solana validator while providing similar rewards and network function.

Again, in terms of PoS chain validators the hardware requirements of Solana are significantly higher than any other PoS chain I'm aware of - it's not even close.

> Solana isn't more centralized because of PoS, it's because it has extremely high validator requirements (including a 300Mbps internet connection, which rules out running one in most homes).

High validator requirements are a barrier to entry, but Solana has a Nakamoto coefficient (decentralization score) of about 30 (actually more, but this is a couple of months ago) which is more than most - probably all - fast chains.

https://youtu.be/5NDs3Il2J3Q?list=RDCMUC9AdQPUe4BdVJ8M9X7wxH...

> but I have no idea why it would lead to less decentralization in practice

Because it takes money to have enough ETH to be a validator and therefore, you centralize. It takes 20ETH to be a validator, and at the height of ETH valuation, that roughly translated to about $100k. Not everyone has that kind of money to stake. Plus, it is well know that majority of ETH or any crypto for that matter is held by a small group of people, which further pushes centralized operations.

As opposed to bitcoin where you need to be able to purchase a large quantity of hardware to have any chance of contributing to validation?

I don’t need 100k to get started.

It takes 32 ETH to be a validator.

Ohh yeah my bad. You’re correct, and it makes it even worse.

I don’t understand the value of Solana. Crypto’s value (at least in part) depends on it being decentralised. I.e trustless. Solana looks like it was relatively centralised, so what’s the point? And why would it be worth much even before FTX collapse?

Crypto's value depends on the perception of decentralization which is more marketing that science. In crypto, "relatively centralized" is the same as "decentralized enough".

In the current take on crypto, sure. Idk if I would call that success though!

The value of crypto to many people was the fact that other people were willing to give them dollars for it.

Seeing as stonks trade for often >40x the company's actual output, sounds pretty similar, doesn't it? Most coins should be worth a fraction of their traded value, and the same with Wall Street. The keyword is speculation that it will eventually catch up to its bullshit traded value. We learned that in 2008, and a lot of people lost their savings and the roofs over their children, then a few years later we completely forgot and now act like crypto invented the idea of trading bullshit.

Sure, it's kind of like really crappy penny stocks that are in danger of being delisted and probably doing fraud. But real companies have cashflows and assets, get audited by real auditors, hire real employees instead of teenage "anons", etc. It's actually pretty different!

Agreed. Shitcoins are like pennny stocks but not even audited or held to much scrutiny at all. It's more like investing in a github repo.

There are, however, plenty of real projects backed by real money with real employees, open books, and have proven resilient over these handful of years. If the economy gets pumped full of fakeass USD causing stonks and crypto to triple up, I'm not going to be one to say "Yeah that value is so real". My fave stonk Square (SQ) (I refuse to call it block), has net me several boatloads of money since day 1, and now it's dropped to only a couple boatloads. Does that mean it's now bullshit? No. It means the economy has a mind of its own and let's not pretend it makes much sense. It's more powered by emotion and guessing than actual worth.

> There are, however, plenty of real projects backed by real money with real employees, open books, and have proven resilient over these handful of years.

Citation needed, especially considering what’s been going on with BNB

Well, right. And real companies don’t just have cheer leaders on YouTube making videos of being on team “X-coin” and having tons of followers like a cult trying artificially prop up the value so they can dump it on sone sucker. It’s really amazing.

> Seeing as stonks trade for often >40x the company's actual output, sounds pretty similar, doesn't it? Most coins should be worth a fraction of their traded value, and the same with Wall Street.

Well for one “40x their output” is meaningless. Ideally stocks are valued at their expected net present value of future dividends; implicitly assuming it’s baked in for growth stocks.

Second, I don’t care at all what you think is overvalued. The question was “why was this thing valued without crypto fundamentals?” And the answer is because the market cares very little for those things.

> I don’t care at all what you think is overvalued

K that's the topic of this conversation.

Read again. The topic of this thread is “why do things have value”.

What is happening in crypto is that libertarian and other anti-“the system” nerds are just rediscovering scams and schemes from the 1800s.

If you don't count Bitcoin that's true.

Bitcoin is used as a tool against French colonial CFA Franc in Africa. A a lifeline to access the outside financial world for citizens of countries under sanction.

https://iharare.com/african-countries-are-no-strangers-to-th...

All of that added up to a trivial number of useful transactions which should be obvious when you consider how few transactions per day the protocol could handle globally and how expensive transaction fees often became.

The fact that some people have a valid use case does not refute my claim

There's no mention of CFA Francs in that article? Also the bitcoin experiment that this article briefly talks about has since become a failure - though that's probably more on the incompetence of the rollout than bitcoin.

>Bitcoin is used as a tool against French colonial CFA Franc in Africa.

That's a new one. Sure there are some crypto enthusiasts/ scammers (those are not necessarily the same people) trying to push some convoluted argument related to the CFA but this is super niche at best. And furthermore, if anyone would actually use crypto in the CFA area, it would be as a mean to circumvent the exchange control and acquire Euros/Dollars. That's not exactly fighting "colonial money", is it ?

the value of solana was that there were bigger fools out there who would buy your coin from you. We have a word for it, I can't remember...ugh.. something related to pharoah...egypt...scheme..

It's really more of a Ponzi scheme than a pyramid scheme, technically.

A Ponzi scheme implies that there is someone who manages your investment though.

A pyramid scheme only implies that the number of investors must keep growing.

Solana now has a higher Nakamoto coefficient (decentralisation score) than the Eth L2s.

Well yea, Ethereum L2s are all just one sequencer ATM.

The point an L2s is they are secured by Ethereum, not be their own consensus. They have an escape hatch so if they ever go down / disappear / attempt to steal all the money - then anyone can force a withdraw to Ethereum and be safe. If Solana goes down there's nothing you can do.

I should probably have also pointed out that Sol has a greater Nakamoto coefficient than eth.

- Some people want a chain half way between Ethereum and AWS. Games items for example don't need all the security of Ethereum.

- 90% of crypto is still speculators who only care about cashing out before the collapse, if Solana lives only a few years that is fine by them.

The more cynical view: many people had no technical need for a chain whatever, but wanted to “rub some crypto” on their project to get some hype-driven cash.

The value of crypto is based on the concept of pump and dump.

It's fast and cheap, had the most active wallets of any blockchain, and when you actually measure it it's more decentralised than all the eth layer 2s. https://youtu.be/5NDs3Il2J3Q?list=RDCMUC9AdQPUe4BdVJ8M9X7wxH...

Facebook’s Libra was in Rust too

A lot of cryptocurrencies are in Rust these days. Half of the crypto stack of Mina is, Sui is, Linera is, Grin is, Penumbra is, etc.

Grin? Haven’t heard about grin in ages. I used to run a node, thinking about bringing it back online.

Grin is that rare blockchain focussed on simplicity and fairness [1].

[1] https://np.reddit.com/r/grincoin/comments/ug70ft/are_there_a...

In fact, good luck finding a Rust job these days that doesn't involve crypto.

to be fair it's pretty surprising you can find rust jobs at all, a luxury most languages lack

Programming languages of top 100 crypto projects: https://github.com/slowernews/notebook/blob/master/on-crypto...

At our company we don't think account is an elegant model. There are many different things called account in Solana. This is a presentation from our Spanish speaking team talking about Solana from a critical developer approach: https://docs.google.com/presentation/d/1icRB_oYx_eIF5ytYTb94...

The slide #6 talks about the different things called account. You can easily translate it if you don't speak Spanish.

Well, the accounting model is not about naming...It is about detecting dependent transactions very early so you can figure out how many can go into a block in parallel. Another one I like is the pricing model, a.k.a rent. You pay the rent, and your account survives in the chain. Once you are done, you stop paying rent, and the runtime sweeps your account under the carpet. (you can also create your account rent-exempt and get all the money back).

You can't run a solana node behind nat, last I looked, because it uses some weird custom udp protocol that doesn't work right behind nat.

That's where my experimentation with the chain stopped.

This is a common problem for any peer to peer protocol.

Doesn't really strike me as a problem though, it just means the client has responsibility of opening a port. Considering the complication of adding 1000s of hole punching strategies in the protocol, it's a fine trade-off.

Not really. Bitcoind and geth work just fine behind NAT. They have no hole punching strategies in the protocol at all.

Hu yes they do. Last I checked they both had NAT traversal through UPnP. I remember that years ago bitcoin even had peer discovery through IRC channels in case UPnP was disabled.

1) UPnP isn't part of the p2p protocol

and

2) even in the case where UPnP fails and there is no inbound connectivity, bitcoind works and syncs fine with only outbound connections

Solana won't even sync behind nat.

I don't think so. I don't believe any Blockchain has as much instability as Solana and security breaches as the Solana ecosystem. Then look at how many projects FTX and Alameda owned or were invested in and all the hacks involving their projects and the questionable auditing companies like OtterSec.

> Then look at how many projects FTX and Alameda owned or were invested in and all the hacks involving their projects and the questionable auditing companies like OtterSec.

I would expect breaches in many Solana projects just as I would in other blockchain projects but after looking at this more in both here [0] and here [1] it is more Ethereum projects that have shown to have these hacks for years (due to it being there longer than other projects).

I have not seen the overall blockchain(s) of Ethereum or Solana getting outright hacked at the core. But Polygon comes to mind as a highly centralized chain which got its whole chain breached [2] requiring the developers with access to the chain to force a silent update on validators.

That doesn't look 'decentralized' or 'secure' to me on the core level of a blockchain in the case of Polygon which another VC and FTX funded chain.

[0] https://web3isgoinggreat.com

[1] https://rekt.news

[2] https://polygon.technology/blog/all-you-need-to-know-about-t...