Just to be clear.
To spend a DBC you need to get it seen by the majority of elders, you also need to provide the signed proof the parent was spent. While that’s happening you cannot spam more tx for that DBC.
As listed above there are many ways to prevent or at least make spamming more expensive for the attacker. However a determinied attacker cna cause trouble and slow down the network, like the BTC dust attacks a while back. But that is not necessarily a network killer.
Why nano crypto (“instant” without fee payments) had problem with spam attackers, it similiar or no?
Hopefully network capacity is so high that only attacks are the problem, but if transactions are free and nodes not incentivised, it may be that legit demand becomes greater than capacity from non-spam transactions (e.g. if people tokenise assets & run popular DEXs, or gaming use cases take off, leading to huge legit transaction volumes).
If legit usage causes the capacity to fill up without incentives for more nodes to join / a fee market to ration transactions, it could become a problem for the operation and growth of the network.
Interesting that this is happening. I can see reasons why you would want this.
From whats been said it sounds like most of the pieces are in place to launch this network. Is that true?
I see that there is some unclarity wrt fees and rewards.
In payment network there are fees on transfers. Those pay the rewards and fill the same function as store cost.
If SNT in data network will have a fee on transfers is not decided. The aim has been to have it free. Naturally there must be a hinder to abuse. If client extra-work is not enough then something more is needed. Whatever the solution will be, it will of course be addressed.
Someone tell me that in maidsafe network there isnt transaction fee, so its untrue for payment network? 
It is the hope and intention to not have transaction fees.
However if the network fails to work without transaction fees, then
there will be transaction fees.
The payment only network is the first step and will have a fee (probably).
BUT, when the data network is added to the payment network, then the transaction fee will go away - assuming the network can function without it.
The developers didn’t learn to code at Hogwarts. They can’t use magic. If a solution apart from a transaction fee doesn’t exist and the network requires transaction fees to function, then that’s what it will have.
Better to have a network with transaction fees than no network at all.
Hopefully there will be no need. Time and testing will tell. I hope that’s clear.
It’s not unreasonable to have fees.
I do like the sound of @Southside’s idea of after a certain number of transactions you could have the client do some PoW. In the ChatGPT list @Knosis provided it mentioned PoW too as an obvious first choice but also mentioned rate limiting, which I think has been suggested here in the past. Also mentioned was something we already have which is small sections that limit overall network overload/exhaustion. 
Perhaps the TX fee, if still necessary with the addition of some of these other inhibitors, could be negligible. Like a micro fee.
I’m sure the team will do the best thing for the stability and health of the network. They never cease to surprise or amaze anyways.
Why not have escalating fees? If an account exceeds some amount of transactions in a given span then fees escalate. I know the concept of “time” is kind of relative in the network, so say the initial fee is something like 0.0001%, or even free. If an account exceeds 1% (these numbers are just made up for example) of the last ~1000 transactions in the section, then their fees escalate to 0.001%. If they reach 10%, then 0.01%.
Thus, if some corporation like a Walmart or Amazon starts accepting SNT, they would likely hit that 0.01% fee, spammers would hit that rate, and your every day users would essentially have no fees.
There aren’t accounts on the Network though, it’s just keys, and there is nothing to stop an attacker creating millions of them.
I didn’t realize Safe auth was gone forever. I thought it was just temporarily removed. Fair enough.
Every so often people suggest using PoW for something and I’m surprised. If we’ve learned anything about this isn’t it that is an all round bad idea unless you really have no alternative? I won’t bother to go through the issues with it, but maybe someone needs to?!
I am a bit scared of you arent sure if u(we) will have fees or not. If yes, will be this still no blockchain project? Isnt this project too much theoretical? I thinked this basic stone of fees or not can be handled on mathematical paper level. But probably not I am amateur so.
If there are better alternatives, absolutely. But I would say it’s not wasted if it is only under the circumstance of excessive transactions. At that point it is like you are working for those transactions on behalf of the network to avoid stressing it. So not entirely without purpose and not anywhere near the level of mining/securing a global ledger. Just enough to be prohibitive en masse. No application specific integrated circuits or heavy GPU’s, just some extra work for large amounts of transactions.
To avoid such work, exchanges or the like would do the same thing they do on blockchain and batch transactions.
As David has said, each client is only able to process one transaction at a time so that prevents a lot. I guess really the issue is with a payment only network that the TX fee (store cost) is more so for nodes resource rewards than spam prevention.
But if we want free TX’s with storage network then maybe just a dash of rate limiting, or PoW, or micro tx fee, could be enough to feel like free and keep the barrier low for all globally to enjoy the security and freedom of transacting on SN.
Nothing you say is wrong. PoW still has numerous reasons to avoid it if at all possible, so I’m surprised that it often seems to be the first thought when there’s a problem of this kind.
My instinct is the opposite. 
Proof of Sloth - ‘Well, I was going to doublespend and all that malarky, but in the end I couldn’t be arsed’.
In that respect PoW is kind of like duct tape. Not exactly a proper fix but what is usually on hand and able to rig it up. Appreciate the call out though. Something more elegant and light would be better but I would also assume it’d be seen as a disadvantageous and so batching or otherwise would be preferable if not malicious.
This is easily gotten around, multiple accounts, multiple clients
But these are unneeded since the network doesn’t know who the client is anyhow. No need to be logged in since these transactions can essentially be done offline.
PoW is also a very unfair method. The phone compared to my 24 core PC for instance, PoW for me would be nothing whereas the phone would take an hour (or day) to do PoW that affects my PC. Also the network doesn’t know what machine I am running on, it only knows the packets sent and a PC could claim to be a phone.
The Lag across the network will slow down individual transactions to maybe a second or two or … to complete. At one second an individual can only do approx 85,000 transactions in a day and at 3 seconds its approx 28,000 transactions a day.
say 5 seconds of PoW means 17,000 transactions a day (PoW could be done in parallel with getting sigs). Now if its 5 seconds on a phone then its milli seconds on my PC so my PC is back to only lag affecting it, and if 5 seconds on my PC then its hours on my phone. And if you try to have different amounts for phone and PC then my PC’s client presents itself as a phone to the network.
I really think PoW is not a suitable solution when you want nigh on real time transactions. Also across a billion clients the energy wasted by PoW might dwarf bitcoin energy wastage.
A better solution is one that requires more lag time getting sigs. Even having the nodes delaying random time between 1 and 5 seconds before responding with the sig would be far better than PoW.
Excellent points I hadn’t considered. Informative reply and always welcome. Thanks Rob!
I was catching onto the workaround based off Jim’s response earlier about endless accounts being able to be created.
So if a malicious actor wanted to spam the network they create tons of accounts and have them all do a bunch of transactions offline and then execute them all at once online, correct?
So instead of PoW, we need PoWoT - Proof of Waste of Time.
Except we don’t have time as such…
I have faith in the team and community to find a solution 