Author: 0xTodd

(*Note: This article is solely @0xTodd's own work, not AI generated.)

I feel that there is a very unhealthy trend right now: either people are overly praising OpenClaw/Lobster to the point of deification, or they completely demean it, reducing it to a mere information boiler room pyramid scheme.

Objectively speaking, I think Lobster can get a score of 65 out of 100.

【Lobster Actually Has Only Three Main Skills】

Lobster essentially consists of: an external brain (Claude, GPT, MiniMax, etc.) + hands-on ability + super long prompts.

So, first of all, you cannot expect a major breakthrough in its intelligence; its intelligence hasn't broken through at all. Its intelligence is 99% dependent on which model you are using.

Of course, I suspect that some people have excessively hyped it up; a major reason for this is that they have never used Claude Opus 4.6 before. Now, having switched to a new model via some intermediary, like OpenRouter using Claude, they feel suddenly smarter.

The second point is hands-on ability.

Previously, these AIs were merely consultants, and no one really likes consultants. Now that they have hands-on ability, they feel a bit more like employees. But honestly, they are only at the level of interns, albeit sweet-talking ones.

The third is prompt words.

Because Lobster comes with a set of super long prompts by default. Honestly, the vast majority of people's prompt writing skills cannot reach that level.

So after using it, everyone feels: “Wow, Lobster is amazing!”

Moreover, humans tend to be lazy; no one likes to come up with a huge long set of prompts during traditional chats.

【Who is Criticizing Lobster?】

However, on the flip side, I can understand the motivations of those who criticize Lobster as being entirely worthless.

Lobster spends money without making any.

Because fundamentally, this is not just a problem for Lobster. Even if you use GPT or Claude, you also won’t make money just by using these AIs.

Forgot about the AI crypto trading competition? Didn't everyone lose out?

PS: The result of the AI crypto trading competition, Doubao won—because Doubao didn’t open an account.

They can't make money in their normal form, it's not that packaging them into Lobster will enable them to earn money, because their intelligence remains unchanged.

Since AI truly cannot make money, some people can only make money through the following means: installation services, selling courses?

If you want to profit from selling courses and installation services, you must hype Lobster to be something miraculous; this has been the approach since the time of Mi Meng.

In that case, it’s just a 65-point product, but you forcefully claim it to be worth 130 points and the revolutionary Web 4.0.

Everybody ends up wasting time, and after using it, realizes: “Wow, it’s not even 130 points.” Naturally, this sense of disappointment will be quite strong. The difference between buyer's showcase vs seller's showcase is huge, making a misfire almost inevitable.

【My Evaluation】

If you ask me for an evaluation, I actually quite like the form of Lobster.

I think it has a certain persona, which I mentioned before, and it feels great.

Moreover, I am currently using Minimax M2.5, which is domestically produced and comes at a very low cost. The image recognition uses Claude Haiku, so it finds a sweet spot.

It chats with me and does some small tasks without costing many Tokens; it’s quite pleasant.

This is also a little horse crossing the river problem. A friend is using Claude Opus 4.6, which is 20 times more expensive than mine, and if the output isn't good, it can be quite frustrating.

I haven’t spent much money; if what I generate isn’t usable, I can just discard it without feeling bad at all.

After all, Vibe Coding is a new era slot machine: I play once for 5 cents, while you play once for 1 dollar; even if my explosion rate is slightly lower, the price difference makes the experience completely different.

【Security Issues Cannot Be Compared to Horse Racing】

Everyone keeps saying that OpenClaw has surpassed React's 240,000 stars accumulated over 13 years in just 3 months, and has even left behind the 40-year-old Linux kernel.

But little do they know that the number of bugs in OpenClaw has also left Linux behind??.

If you take a close look at the OpenClaw page on GitHub, that star count is indeed in the top 10 on the entire internet. The various bugs in the issues have also reached an astonishing 5000+.

What does 5000+ mean? Is it because GitHub projects can only show a maximum of 5000+ bug issues?

This is not OpenClaw's limit; this is actually GitHub's upper limit.

In comparison, another open-source software with 200,000 stars, Vue, only has 359 issues.

So I think many ordinary people are actually not suitable for using open-source software. Open-source software is best for those with a certain foundation, especially those who have a sense of security.

If you recklessly give OpenClaw a lot of permissions, I think in the end, for example, it could get hacked, right?

(Actually, there aren't many who get truly hacked, but such news is very shareable. One moment they say tens of thousands of people are exposed, and the next moment they mention credit cards being maxed out, which can naturally create negative perceptions about Lobster.)

At such times, you cannot demand the level of mature, security-verified software aimed at a wide audience like Gemini, Claude, and Doubao or Qianwen from an open-source geek software.

This is like comparing horses in a race.

There is a saying in the software development community called “anti-stupidity mechanism.” Those traditional AIs have done well in preventing stupidity, but over here at Lobster, there is currently a pile of bugs. After all, it is open-source software; it was even written as vibe coding open-source software.

Even Google had a news segment recently, saying that a man in Florida had been chatting with Gemini for a long time and subsequently committed suicide; his family even sued Google.

Just think, if a big company with such high security standards can still encounter issues, then you can’t mention Lobster, right?

【Conclusion: Should I Install Lobster?】

So in summary, ultimately, if you ask me to objectively evaluate Lobster, here’s what I think:

If you have some technical foundation and enjoy tinkering, then treat it as a research project.

Or, if you simply have some free time and want to earn the eggs distributed by Shenzhen.

I think it’s also fun as a pastime, just approach playing with Lobster with this mindset.

But don’t have the mindset of “making big money” or “leading Web 4.0,” as that will definitely lead to disappointment.

If you are usually busy and don’t have time to tinker, then use the regular Claude, Gemini, or even Doubao, DeepSeek; I think you won’t miss out too much on this so-called revolution.

The term FOMO is originally about fear of missing out. But I have used this for several days and can tell you clearly, you won’t miss out on too much.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。